Garuda 200

Indonesia’s National Transport Safety Committee (NTSC) recently released a final report on the accident which has been made available by our Australian Transport Safety Bureau (ATSB). Its thoroughness is in marked contrast to the NTSC report, produced under former leadership 10 years previously, on the “accident” to a SilkAir Boeing 737 which dived into the Musi River in Sumatra in December 1997.

The Indonesian NTSC is to be commended on the detail of its full report, which can be downloaded from the ATSB’s website.

Anatomy Of An Accident

Yogyakarta, near the southern coast of Java, is slightly less than 450km east of Jakarta, and the early morning domestic flight from Jakarta in fine weather on March 7 2007 was expected to take well under an hour. On board Garuda Indonesia Flight 200’s Boeing 737-400 PK-GZC were 133 passengers and a crew of seven.

Taking off from Jakarta at 6. 17am local time, the flight continued as planned at Flight Level 270 (27,000ft). As the aircraft was nearing its destination just after 6. 45am, still in cruising flight, Jakarta ATC instructed the crew to “maintain level 270 and contact Yogya Approach 123. 4”.

Twelve minutes later, Yogya Approach cleared them to Yogyakarta via airway W17 for the airport’s Runway 09, with a requirement to report leaving FL 270.

Acknowledging the clearance, the first officer sought confirmation that they were cleared to descend to circuit altitude. Approach replied “descend to 2500 initially”. The crew began the descent and the captain, who was flying the aircraft, briefed the first officer on the approach procedure, including the ILS and the possibility of a localiser approach in the event of glide path failure. But he did not include any visual approach briefing.

As they passed through 7000ft, the controller asked if they were operating in visual conditions.

Told that they were, he cleared them for a visual approach to Runway 09, telling them to “establish the aircraft on a long final” and to report “runway in sight”. But the captain simply continued the ILS approach, without advising the controller they were doing so.

The wind was westerly at nearly 30kt, and now below 6000ft, the captain apparently realised they were excessively high for the near straight-in approach to Runway 09. He

Obviously still much too high, the captain again pushed the nose down steeply in an attempt to reach the runway.

steepened the descent, increasing the airspeed from 288kt to 293kt. It then reduced to 243kt. Garuda’s Operations Manual specifies the maximum speed in terminal areas below 10,000ft as 250kt.

Ten miles from the runway at 6. 55am, the Boeing passed over the Runway 09 ILS initial approach fix nearly 1500ft higher than the published height of 2500ft, at an airspeed of more than 280kt. The first officer called the Tower and the controller transmitted, “Continue approach, Runway 09, report final”.

The captain read back the landing clearance.

On the ground, the wind was moderate, but gusting at times to between 20 and 30kt.

With only one degree of flap extended, the aircraft crossed the final approach fix still nearly 1000ft too high, and the captain called for the undercarriage. Half a minute later he called, “Check speed, flaps 15”. The airspeed had reduced to 238kt but was still excessive for this flap setting (maximum 205kt), and the first officer instead called “flaps five”.

The Tower now transmitted, “Check gear down and locked, clear to land Runway 09”.

Obviously still much too high, the captain again pushed the nose down steeply in an attempt to reach the runway. As he did so, the airspeed increased excessively again, the rate of descent rising to almost 3500ft per minute, and the ground proximity warning system (GPWS) began sounding “SINK RATE” alerts. Then, less than 1000ft above airport elevation, with the airspeed fluctuating between 230 and 250kt, the GPWS twice sounded “TOO LOW – TERRAIN”. The first officer urged the captain to go-around.

But the captain acknowledged the Tower’s landing clearance, and again called to the first officer: “Check speed, flaps 15”. The rate of descent had decreased to 1600ft/min, but at less than 600ft above the ground, the airspeed was still more than 250kt. Ten seconds later, when the aircraft was only 150ft up at 240kt, the GPWS sounded “WHOOP, WHOOP, PULL UP” twice. Alarmed, the first officer called out “Oh Captain! Go-around Captain!” Disregarding both the GPWS warnings and the first officer’s call, the captain merely said, “Landing checklist completed, right?” persisting with the fast approach with only five degrees of flap. He then repeated his call for ‘flaps 15’ three more times. But because the airspeed was still far too high, the first officer left the flap selector at the five degree setting.

The Boeing finally attained the glideslope as it neared the runway threshold, crossing it at about a height of 90ft at an airspeed in excess of 230kt, nearly 100kt faster than for a normal landing at ‘flaps 40’. With the surface wind northeasterly at nine knots, the ground speed was 235kt. The aircraft touched down heavily about 860 metres beyond the threshold at an airspeed of 221kt, 87kt faster than the landing speed for 40 degrees of flap.

In a shrill voice, the first officer cried out again, “Go around!” But again the captain disregarded him.

The Boeing bounced twice, each runway contact becoming heavier, with the third peaking at 2. 91g. The first contacts were on the main wheels, but the third, with the aircraft about one degree nose-down, was on the nose leg. A nosewheel tyre burst as the leg touched down hard after the third bounce, sparks flying up as the wheel scraped the runway, and the wheel assembly broke off. Both thrust reversers were deployed, but were stowed seven seconds later when the captain shut down the engines before the aircraft overran the end of the runway, still travelling at 110kt.

About 160 metres further on it crossed a road, hit an embankment and the engines struck a concrete gutter. The engines and undercarriage were torn off, and the cockpit was pushed back over the forward cabin. At the same time, the starboard wing was severed from the fuselage. Its fuel tanks exploded, hurling the wing over the fuselage on to the top of the port wing. Fire broke out immediately, fed by fuel from the disrupted wing, progressively destroying much of the aircraft structure. One flight attendant and 20 passengers suffered fatal injuries and burns, and one flight attendant and 11 passengers were seriously injured.

Airport fire-fighting personnel, standing by as the Boeing was on final approach, saw it landing unusually high and fast, and realised that an overrun accident was imminent. Two fire trucks set off immediately, but were forced to stop at the perimeter fence about 160 metres beyond the end of the runway. Lack of an access road and uneven ground prevented them reaching the burning aircraft, and it was too far to spray foam from where they were. Moving

With the airspeed fluctuating between 230 and 250kt, the GPWS twice sounded “TOO LOW – TERRAIN”

fast, they rigged a flexible extension hose in about 15 minutes, but the whole operation then became uncoordinated.

The hose was punctured in a number of places by other rescue and onlookers’ vehicles driving over it, the leaks rendering its pressure too weak to be effective, and it was unable to blanket the whole wreckage. Bystanders added to the confusion by all trying to tell the fire-fighters what to do, effectively preventing the fire chief directing the operation.

Two city fire-fighting vehicles arrived about 45 minutes later, but they carried only water. More confusion reigned when someone, untrained in fighting fuel-fed fires, directed them to start hosing the fire. Altogether, it took more than two hours to extinguish the fire.

Impact forces and fire were most severe in the forward section and the starboard side of the aircraft, where the majority of the passenger victims were seated. One flight attendant, in a crew seat in the forward galley area ahead of the business class section, received fatal burns, and one other flight attendant was seriously injured. Neither pilot was injured.

The remaining flight attendants helped passengers to evacuate the cabin through the rear door and over-wing exits on the port side of the aircraft, and rendered assistance to the injured while they waited for ambulances and medical assistance to arrive. The whole rescue operation took until late afternoon.

Investigation

Badly damaged, the aircraft came to rest in an open rice paddy 250 metres from the end of the runway, close to the approach lighting structure. Damage to the engines’ fan blades was consistent with low RPM at impact. The starboard side of the fuselage was destroyed by fire, with all structural damage the result of excessive structural overload during the impact sequence. The port rear door slide was out of its container with the slide bar attached to the door frame. No other evacuation slides were deployed, but the port over-wing emergency exits were open.

The airport – Yogyakarta’s Adi Sucipto Airport is a joint military/civil aerodrome, 350ft above mean sea level. Its single paved runway is 2200 metres long, aligned 09-27. At each end of the runway there are paved end safety areas. The area at the end of Runway 09 at the time was 60 meters long, with an additional grassed area 98 metres long. This did not conform to ICAO standards which specify safety areas shall extend from the end of a runway to at least 90 metres.

The airport’s control tower is operated by military controllers holding both military and civilian licences. No aircraft operating into Yogyakarta on the day of the accident reported any abnormalities with the ILS, and all communications between ATC and the aircraft were of good quality.

The aircraft – PK–GZC, a Boeing 737-497, was 17 years old and had flown 35,200 hours. Its weight at the time of the accident was below the maximum for landing, and within centre of gravity limitations.

It had been correctly maintained, was fully serviceable when the accident flight began, and there was no evidence of any defect or malfunction that could have contributed to the accident. All inflight communications between ATC and the crew were of good quality, and radio transmissions did not indicate any aircraft defects.

Flightcrew – The captain, 45, was appropriately licensed and rated. He had logged 13,400 flying hours, of which 3700 were as pilot in command on 737s.

He had flown 90 hours during the last 28 days. His last flight into Yogyakarta was a week before on February 28 2007 when he landed on Runway 27. His previous most recent landing on Runway 09 was two weeks before.

The captain had attended a GPWS introductory seminar in 2005, but had not been checked or given simulator training in vital actions and responses to GPWS alerts and warnings.

During a simulator proficiency check in September 2005, a check pilot noted his “speed high on touchdown” during single engine landings. During a line check in August 2006, another check pilot noted that he “did not comply with speed restriction procedure” during arrival.

A check pilot who conducted a further simulator proficiency check in September 2006 did not express a concern about the previous speed findings.

The first officer, 31, held a commercial pilot licence. He held an endorsement for the Boeing 737, a multi-engine instrument rating and had logged more than 1500 hours, of which 1350 hours were on 737s. He too had attended a GPWS introductory seminar in October 2005, but there was no evidence that he had been checked or received simulator training in responses to GPWS alerts and warnings. His last flight into Yogyakarta was three weeks before.

Weather – The area forecast available to the crew at Jakarta showed no significant weather. The en route wind was expected to be westerly at 10 to 15kt at FL 270 and 35kt at FL 280.

Flight recorders – The flight data recorder (FDR) and cockpit voice recorder (CVR) were recovered from the accident site, and taken to the ATSB’s replay and analysis laboratory in Canberra. The FDR was severely burnt, but the crash protection module had survived and contained data relating to the entire accident flight, as well as 31 previous flights.

The CVR recorder sustained heat damage but appeared to have survived the impact and fire. However, normal recovery methods were unsuccessful in downloading its data. The manufacturer was consulted, but prescribed methods of recovery also failed to recover data.

The ATSB then sent the CVR to the manufacturer in the USA under the supervision of an ATSB investigator.

Using specialist equipment, the manufacturer found the pointers in the electrically erasable memory were corrupted. This had probably occurred during the severe impact sequence.

Resetting the pointers could be accomplished only at the manufacturer’s facility, and when this was done, the data was successfully downloaded. The data provided valuable flight information which, in conjunction with that from the FDR, enabled the entire flight to be reconstructed in great detail.

Excessive speed – During the approach to Yogyakarta, the captain flew the aircraft at an excessively steep flightpath and high airspeed, resulting in an unstabilised approach. He did not follow a company requirement for a stabilised approach, and did not abandon the landing to go around when the approach was not stabilised. With his attention fixated on landing, he either did not hear, or disregarded, the GPWS alerts and warnings as well as calls from the first officer to go around.

The captain told investigators he continued to call for flap 15 because he was committed to land from the approach.

He was aware he would not be able to use flaps 40 as planned, but believed he could safely land using flaps 15, even with the higher airspeed required.

The first officer said he did not extend the flaps to 15 degrees as instructed, because the airspeed exceeded the maximum operating speed for flaps 15.

The captain argued that he was unaware of the actual airspeed, and expected the first officer to inform him of any speed concerns.

Garuda’s Operations Manual instructs a first officer to take control and execute a go-around when an unsafe condition exists. Although the first officer did not follow this direction and take control when the captain repeatedly ignored the GPWS alerts and warnings, there was no record of his having received simulator training in the vital actions required to retrieve a situation that could compromise the safety of the aircraft. Garuda’s simulator records likewise showed no evidence of training in vital actions and responses to be taken in the event of GPWS alerts and warnings.

Garuda was required to comply with all Directorate General of Civil Aviation requirements and DGCA would conduct ‘complementary’ surveillance. But between 1998 and July 2007, surveillance had been conducted only on two occasions.

Deficiencies in Garuda’s training and checking procedures for pilot response to GPWS alerts and warnings, and procedures for a first officer to take over in the event of a captain flying unsafely or becoming incapacitated, apparently went unnoticed by the DGCA.

Analysis

The captain and first officer were communicating effectively during the cruise and initial descent phases of the flight.

When the captain realised the aircraft was above the planned profile, he tried to regain it by losing more than 4000ft.

Over a two minute period the speed increased excessively, but this achieved a loss of slightly less than 2000ft.

The aircraft was then below 10,000ft at a speed above the 250kt required in terminal areas. Yet the speed brake was not deployed and the crew did not seek ATC approval to exceed 250kt.

When passing through 4000ft, the captain said in Indonesian, “Oops, strong wind”. He had apparently not noticed the strong wind at higher altitudes, the wind having actually decreased at lower altitudes.

The first officer did not provide effective monitoring and operational support to the captain, showing no evidence that he was aware of the strength of the tailwind and its effect on the flight.

The captain subsequently assessed the situation by calculating the altitude and the remaining distance to the runway, and decided that the flight profile was not as he had expected. He then attempted to lose height by steepening the descent, but succeeded only in flying a flightpath that was erratic in pitch, causing the airspeed and altitude to vary considerably, and resulting in an unstabilised approach.

Garuda’s Operations Manual required the aircraft to be configured for landing, with undercarriage extended, flaps 15, and airspeed at 150kt, when approaching the final approach point. But when it passed this point, the speed was 254kt (ground speed 286kt), and it was still in a clean configuration.

Realising the situation, the captain extended the flaps to the one degree position with the aircraft nearly 1200ft above the approach profile for the ILS.

Although the crew had been cleared for a visual approach, they executed the ILS approach, but did not inform the Approach controller.

The captain again expressed concern that the vertical flightpath was not proceeding normally, when just before 6:57am he commented “Oh there is something not right”. Between this time and a little after 6:57 the airspeed was varying between 229 and 244kt, pitch was varying between 3. 5 degrees up and 3. 8 degrees down, and the rate of descent reached 3520ft/min, the captain’s attempt to regain the glideslope resulting in an excessively steep approach path. Garuda’s standard operating procedures state that to achieve a stabilised approach, “only small changes in heading and pitch are required”.

The Operations Manual states that during the descent phase, at altitudes below 10,000ft, and during taxiing, all cockpit crew members shall concentrate on cockpit procedures, adhere to monitoring or lookout procedures, and refrain from any non-essential activities.

During the flight below 10,000ft, the captain was not following these directives.

He was no longer situationally aware, and rather than flying the aircraft attentively, he was “focused on landing the aircraft”, as he told investigators.

The first officer did not select flaps 15 when instructed, and, after receiving six GPWS sink rate alerts, he read back the aircraft configuration of “flaps 5”.

He did not tell the captain his reason was because their airspeed of 240kt far exceeded the flaps 15 maximum. The captain did not reduce the speed to the target of 141kt for the approach, the actual speed being 245kt. The aircraft was not in the landing configuration, and its sink rate of 3500ft/min exceeded the Operations Manual requirement of not more than 1000ft/min. As well, the landing checklist was not completed.

The manual instructs that “An approach that becomes unstabilised below 1000 feet in IMC or below 500ft in VMC requires an immediate go-around. ” The aircraft was never in a stabilised approach, so a go-around should have begun at once. Furthermore, the GPWS sink rate alert sounded several times during the approach, as did the “too low – terrain” alert.

When the aircraft was down to 217ft above the airport elevation, still excessively fast, the first officer exclaimed, “Oh Captain, go around Captain!” Even so, the captain continued the approach.

Even another GPWS “WHOOP, WHOOP, PULL UP” warning, and another alarmed ‘go-around’ exclamation from the first officer, did not alter the captain’s determination. The GPWS warnings alone should have given the pilots a clear indication that the approach was unstabilised. The captain did not seem to have the same appreciation of the risk he was taking as the first officer, and his intention to continue the landing was only reinforced when he asked the first officer if the landing checklist had been completed. But the first officer did not respond.

The normal flap setting for a landing is ‘flaps 30’ or ‘flaps 40’, and the captain had briefed the first officer for a ‘flaps 40’ landing. The captain told investigators he had decided to land the aircraft from the approach, despite it being in an unstabilised condition, not being configured for a landing, the GPWS sounding alerts and warnings and the first officer not confirming the completion of the landing checklist.

Seventeen seconds before touchdown, the first officer instructed the cabin crew, “flight attendants, landing position”. This was insufficient time for them to comply with the company requirement to sit quietly for a minute to recall emergency memory items. However, the flight attendants were able to sit and fasten their seat belts before the landing.

The captain’s actions were consistent with being trapped in the condition called ‘fixated on one task’. But because the aircraft was not configured for landing, and still 98kt too fast as it crossed the runway threshold, a go-around was the only safe option still open to him. The touch-down should have occurred around 300 metres in from the threshold, but the aircraft actually landed about 860 metres from the threshold.

The captain told an investigator “his goal was to reach the runway and avoid severe damage”. He “heard, but did not listen to the other voice (the GPWS), and flaps 15 and speed 205 was enough to land”.

Apparently experiencing a great sense of urgency, he sought to avoid what he saw to be a looming catastrophe, fixating on what “seemed most obvious” by aiming to get the aircraft on the ground by making a steep descent. As events proved, his decision was the dangerous option.

Available rescue and fire fighting services at the airport did not meet ICAO requirements. Vehicles were unable to traverse the terrain between the end of the runway and the swampy rice field.

And lack of coordination at the crash site resulted in ineffective fire fighting, with numerous unqualified people interfering in the fire chief ’s role. Had the fire service been equipped to ICAO standards, injuries might have been mitigated and survival prospects improved.

Summary

• Deviations from recommended practice and standard operating procedures are a potential hazard, particularly during approach and landing, and increase the risk of accidents.

• Crew coordination is less than effective if crew members do not work together as an integrated team. Crew members have a responsibility to ensure the safety of a flight is not compromised, by complying with standard operating procedures.

• Support from regulators and all levels of airline management, procedures, training and checking, and oversight and surveillance, is essential to ensure pilots are appropriately trained and equipped for safe flying operations.

An Australian Comment

The scope and openness of the report into this accident provides plenty of sobering food for thought for pilots, airline management, airport operators and aviation fire-fighting rescue services alike.

“Worthwhile reading, if only for the almost totally unbelievable lack of airmanship by the captain and the weakness and ineffectualness of the first officer,” remarked Captain John Laming, a highly experienced international Boeing 737 pilot and a seasoned simulator instructor. “Over the fence at 232kt with flap five (100kt too fast) is the stuff of nightmares. ” “The point made of lack of GPWS training in the simulator is important. It applies to many airlines, not only Garuda.

Full marks to our ATSB for making the report available in Australia. ” Lets hope the Indonesian and international aviation industries learn the lessons available to them from such a comprehensive accident investigation report.