Factual information

The F-27 aircraft had been carrying out circuit training and had landed on runway 05. The crew accepted the Aerodrome Controller's (ADC) offer to backtrack on runway 05 for further circuits and they reported ready for takeoff at 1002.32 CST. The TB-10 aircraft was making an approach to runway 12 for a touch-and-go landing. At 1003.05, the ADC cleared the aircraft for the touch-and-go and also passed departure instructions to that aircraft. Co-ordination between the ADC and Approach controller resulted in an agreement to sequence the F-27 so that it would land behind four jet aircraft arriving for runway 05. The ADC informed the F-27 crew of the sequencing difficulty and, at 1003.15, said there would be about a 30-second delay before he could issue a take-off clearance. At 1003.45 the pilot of the B737 contacted the ADC whilst on approach for runway 05. At 1004.58, the ADC cleared the F-27 for takeoff.

The crew elected to hold position as they had seen the TB-10 on its approach to land but had not seen it depart after the touch-and-go. From the position of the F-27 part of runway 12 was hidden from view behind buildings. Consequently, the F-27 was held by the crew with power on and ready to commence the takeoff once the TB-10 was clear of the intersection. The B737 crew saw the F-27 lined up on runway 05 and slowed their aircraft lowering, 40 degrees of flap to assist in the sequencing process. At 1005.04 the ADC informed the B737 crew to expect a late landing clearance. At 1005.12 the F-27 commenced its take-off run approximately two minutes after receiving the advice of the expected delay of 30 seconds. The TB-10 was airborne but still not clear of the runway intersection and at this time, the B737 was on short final. At 1005.30, as the F-27 was becoming airborne the ADC cleared the B737 to land. At the same time the B737 crew informed the ADC that they were commencing a go-around which they had started just prior to receiving landing clearance. The go-around decision was based on doubts that sufficient separation existed between the B737 and F-27, especially if the F-27 was to reject its takeoff. The B737's go-around was flatter than the normal climb profile so that the captain could maintain visual contact with the F-27 and to ensure a safe manoeuvring speed. At 1005.50, the ADC instructed the F-27 crew to maintain 500 ft and to turn left onto a heading of 300 degrees. As the F-27 was commencing its left turn, the B737 was turned right about 20 degrees to ensure continued sighting of the F-27.

Analysis

Aerodrome Controller

The ADC elected to process the F-27 by backtracking it on runway 05. The sequence chosen by the ADC was time-critical as it depended on the TB-10 crossing the intersection before the F-27 could take off, and the B737 was dependent on both these aircraft departing before it could land.

The controller realised that the sequence would be tight but he did not consider an alternative plan even though several options were available. Although after line up the F-27 crew were told to expect about a 30-second delay the actual delay was one minute and 43 seconds. When the crew of the F-27 were issued with their take-off clearance the TB-10 was still not clear of the runway intersection which led to a considerable erosion of the time base on which the ADC had based his original plan. Radar analysis indicates that the take-off clearance was issued to the F-27 approximately 20 seconds before the TB-10 had cleared the runway intersection. Paragraph 23 of the Manual of Air Traffic Services (MATS) 6-3-4 specifies the separation standard applicable to a situation in which two aircraft are departing from different runways stating that aircraft B [F-27] shall not be permitted to commence takeoff until aircraft A [TB-10] has crossed the intersection. Paragraph 26 of MATS 6-2-3 allows the controller to issue a take-off clearance prior to the prescribed separation existing if in the opinion of the controller no collision risk exists, and there is reasonable assurance that separation will exist when the aircraft commences its take-off roll. In this case, as the F-27 crew were expecting an immediate departure on receipt of take-off clearance, the ADC could not have reasonably expected the aircraft to delay its take-off roll. The investigation found that no uniformly accepted interpretation of the meaning of paragraph 26 existed between ATS management and some operational staff.

The B737 crew were told to expect a late landing clearance and the ADC issued this clearance between 1005.31 and 1005.33. Radar analysis indicates that the B737 was either just short of, or passing the runway threshold when the landing clearance was given. The F-27 became airborne at 1005.35. Paragraph 32 of MATS 6-3-5 states that aircraft B [B737] shall not be permitted to cross the runway threshold until aircraft A [F-27] (less than 136,000 kg MTOW) is airborne after takeoff. If the B737 was short of the runway threshold at 1005.35, then the clearance complied with the MATS instruction. If however, the B737 was passed the runway threshold at that time, the runway separation standard would have been breached. The ADC believed that had the B737 landed, the separation standard would have been achieved and that the F-27 would be airborne prior to the B737 crossing the threshold. Radar analysis could not determine if the separation standard had been achieved. The standard requires that an aircraft below 136,000 kg MTOW should be airborne before a landing aircraft of any size crosses the runway threshold. Therefore, as an example, an aircraft could be airborne at 70 kts and 500 ft along the runway when a landing jet aircraft crosses the threshold at up to 150 kts. The separation standard would then be breached if an unexpected event forced the departing aircraft to discontinue takeoff and reoccupy the runway. An ADC is required to take into account any unexpected manoeuvres by aircraft when issuing a clearance. In this occurrence, there was a 2-4 second period in which the standard may have been achieved. The B737 crew doubted that the standard would be achieved. Once the B737 had commenced its go-around, the ADC expected the climb to be similar to a normal B737 departure and believed this would solve any separation problems that may arise.

However, the climb profile was more shallow than expected, and the ADC considered further action was required to guarantee separation. Instructions were issued to both crews to ensure that vertical and lateral separation were maintained. Crew of the F-27 After lining up and holding on runway 05, the F-27 crew were given traffic information on the TB-10 which was on final approach for a touch-and-go on runway 12. They sighted the aircraft and monitored its descent and touchdown. The ADC had asked the F-27 crew to be ready for an immediate departure as soon as the TB-10 was airborne. When the take-off clearance was issued, they increased power but held position until they had sighted the TB-10 and this delayed the commencement of their takeoff for approximately 12 seconds. Crew of the B737 When on final approach for runway 05, the B737 crew became aware of the F-27 occupying the runway and aircraft speed was reduced as much as possible. On short final, the ADC issued a late landing clearance expectation and the crew briefed for a possible go-around. The B737 captain was concerned that if the F-27 rejected its takeoff or had a malfunction shortly after lift-off and landed again, there would be insufficient room for the B737 to land safely. As the B737 approached the threshold, the crew could see that the F-27 was still not airborne and well within the distance they would need to land and they initiated a go-around about the same time as they received a landing clearance. To regain safe climb speed and maintain visual contact with the F-27, the captain elected to climb at a sufficiently nose-down attitude until his aircraft was safely clear. The captain considered that the F-27 was sufficiently close to his aircraft that a risk of collision existed if he lost sight of that aircraft.

Findings

1. The ADC misjudged the time required for the TB-10 to clear the runway intersection.

2. The ADC issued a take-off clearance to the F-27 when the required runway separation standards were not met. (MATS provides the ADC with this discretion.)

3. The discretion shown by the ADC in finding 2 above, was inappropriate.

4. The F-27 crew could not see the TB-10 during the ground roll portion of its touch-and-go landing on runway 12.

5. The F-27 crew had good reason not to commence takeoff at the time the ADC issued the clearance.

6. The ADC cleared the B737 to land in the belief that the appropriate runway separation standard would be achieved.

7. The B737 crew executed a go-around manoeuvre when they judged that the runway may not be available for the safe operation of their aircraft.

8. At the time of the B737 go-around, it was unclear (to the investigation team) if the runway separation standard would have been maintained.

9. The B737 climbed at a more shallow angle than that expected by the ADC.

SIGNIFICANT FACTORS

1. The ADC placed undue pressure on himself by initiating a backtrack by the F-27.

2. The decision by the ADC to delay the F-27 until the TB-10 had completed its touch-and-go placed the three aircraft in a position whereby they each required the use of the runway intersection within a very short period of time.

3. The runway separation standard applicable in this case is such that it allowed the ADC to issue clearances whereby the B737 crew were unable to guarantee the safe landing of their aircraft.

4. The ADC did not have a fall-back plan for the sequence he initiated.

SAFETY ACTION

As a result of the investigation the following local action was taken. Air Traffic Services management clarified the interpretation of MATS 6-2-3 para. 26 to the satisfaction of the investigation team and ensured that all Adelaide Tower staff were in no doubt as to their responsibilities in this regard. Additionally, the Bureau of Air Safety Investigation issued the following interim recommendation to Airservices Australia on 29 May 1996.

IR960046 The Bureau of Air Safety Investigation recommends that Airservices Australia, in conjunction with the Civil Aviation Safety Authority, reviews the runway separation standards for landing aircraft in relation to departing aircraft. This review should consider what separation standards may be required to cover contingencies where slower departing aircraft may impede the safe passage of faster landing aircraft. Airservices Australia responded on 29 July 1996. IR 960046 Airservices, in conjunction with CASA, is currently reviewing runway separation standards for landing aircraft in relation to departing aircraft. This review includes an examination of the ICAO standards in relation to departing aircraft, as well as current international practices. The review will also attempt to assess the impact and safety benefit of the introduction of more restrictive practices on runway capacity at specific locations.

It is intended that the review be completed by the end of August.

Response Status: Closed - Accepted.

The Civil Aviation Safety Authority responded on 17 June 1996. In accordance with IR 960046 the separation standards in relation to a faster landing aircraft behind a slower departing aircraft will be reviewed. (Note: the Report states that there is a 2 - 4 second period where the standard may or may not have been achieved: were the tolerances of the radar taken into account in making that assessment?)

Response Status: Closed - Accepted BASI Note: The radar tolerances were taken into account in the investigation.

The pilot reported that on arrival in the circuit area he observed that the employee responsible for inspecting the strip was in attendance, and there were no indications that the strip was unserviceable. After a normal approach and landing with 40 degrees of flap set, the nose was being held off when a bump was felt causing the pilot to believe that a failure had occurred with the nose gear. The pilot continued to hold the nose off until elevator control was lost. The nose then settled, and the propeller struck the runway. The nose gear had failed after striking a hole made by wild pigs, and the hole been obscured by weed growth. A further three holes that were not readily visible were found, but they were not in the landing path.

As the aircraft was levelling at FL390 the left engine appeared to suffer a compressor stall. The crew reduced engine power to idle as a precaution. The engine was running normally at idle therefore a descent clearance was requested and the aircraft continued at a lower altitude to Melbourne for an uneventful landing.

The engine was removed after a borescope inspection revealed damage to the compressor. The engine was disassembled. The damage was found to have caused by the liberation of the inner shroud from one 8th stage compressor stator vane segment. Secondary damage had occurred to both the 7th and 8th stages of the compressor.

The liberation of compressor stator inner shrouds has previously been recorded by overseas operators and is under active investigation by the manufacturer.

FACTUAL INFORMATION

Sequence of Events

During a scheduled transit stop, the number 1 engine required three engine starts for the rectification of a minor oil leak and troubleshooting of other defects before the aircraft was subsequently released for service. Soon after departure, whilst climbing through 8,500 ft at 263 kts, the pilot in command noticed that the number 1 engine start valve OPEN light was illuminated and he requested the engine shutdown checklist. Almost immediately, whilst disconnecting the auto-throttle, there was a bang, a slight yaw to the left and vibration for a very brief duration. During the engine shutdown sequence, the engine fire detection light on the centre panel illuminated momentarily. After securing the engine, 30 tonnes of fuel was jettisoned before the aircraft returned to Sydney for an uneventful landing.

Damage to the Aircraft

Inspection on the ground revealed that only the top sections of the fan cowls, containing the hinge fittings, remained on the number 1 engine. The forward section of the pylon, and the wing leading edge outboard of the pylon, were holed and dented. The paint on the starter case was blistered and there were detached turbine blades in the starter exit screen.

Additional Information

After repairs to the aircraft were completed, a fuel leak was detected from a loose connection at the highest point in the engine environmental drain system. During engine run testing, the leak only became apparent after the third engine shutdown and resulted in the leaked fuel pooling on top of the engine gearbox, immediately above the starter.

ANALYSIS

Examination of the fan cowl latches revealed they had all failed in tension overload. This, together with the nature of the deformations found on the recovered cowl sections, indicated that an explosion had occurred within the cowled area in the vicinity of the starter. The explosion was of sufficient magnitude to deform and weaken the integrity of the cowling in the area of the starter pressure relief panel and to deflect it into the airflow. The high-speed airflow then tore the cowls from the engine. Damage to the pylon and wing leading edge resulted from collision with the separated cowl sections in the airflow.

It is considered likely that fuel had leaked from the environmental drain system as a result of the four engine shutdowns during the transit stopover prior to the flight. Fuel pooled on the gearbox above the starter and evaporated as the unvented under cowl temperature rose during the climb. When the start valve opened, the starter oversped and disintegrated, liberating hot debris in the immediate vicinity which ignited the explosive fuel/air mixture. The force of the explosion deformed the cowl to such an extent that the airstream tore the cowlings apart.

Investigation of the start valve and its associated wiring and control circuitry failed to determine a reason for the uncommanded opening. The aircraft maintenance records revealed a history of starter problems and failures involving uncommanded start valve openings in flight on both number 1 and 2 engines.

There is other documented evidence to show that uncommanded opening of the start valve in flight is not an uncommon event.

SIGNIFICANT FACTORS

The following significant factors were identified as contributing to the accident.

1. A fuel leak occurred from a loose connection in the environmental drain system of the number 1 engine which pooled in the area above the starter.
2. The fuel evaporated to form an explosive mixture in a contained, unvented area within the fan cowls during flight.
3. The engine start valve opened, uncommanded, in flight.
4. The starter oversped and disintegrated, having no engine load to contain its speed.
5. The hot liberated debris from the starter ignited the explosive fuel-air mixture.
6. The subsequent explosive force deformed the cowlings into the high-speed airstream, resulting in overload failure of the cowl latches and loss of the cowl panels.

SAFETY ACTION

As a result of the investigation the Bureau made Interim Recommendation 960026 to the Civil Aviation Safety Authority:

The Bureau of Air Safety Investigation recommends that the Civil Aviation Safety Authority bring this occurrence to the attention of the Federal Aviation Administration and the equipment manufacturers. In the light of the history of this problem, these organisations should consider further research into the possible causes of uncommanded start valve openings in flight.

Similarly, the development of an appropriate engineering modification to the relevant start valve should be considered. This would ensure that if a start valve opened in flight, and a starter overspeed occurred, the potential for an ignition source is minimised.

The Civil Aviation Safety Authority responded on 17 July 1996 as follows:

'I refer to Interim Recommendation IR960026 regarding the incident involving US registered Boeing 747 200, N942PR on 21 April 1995 which recommended that CASA notify both the Federal Aviation Administration and Boeing Aircraft Company of the incident.

I have attached copies of the Authority's correspondence with the FAA and Boeing for your information. In each case a copy of the Interim Recommendation was attached to the facsimile.

The Authority will keep you informed of any relevant responses from either the FAA or Boeing.'

The strip direction at Riddell is 15/33. The intention was to fly circuits. There was a northerly wind blowing so take-off was on the 330 degree strip. A storm was approaching from the south and during the circuit the pilot listened to the Essendon ATIS which indicated that the wind was going around to the south.

On final approach the pilot noted that the windsock was indicating that the wind had swung around the southwest and was about 12 to 15 knots. He had flown a fairly tight circuit. Carburettor heat was selected on base when power was reduced and deselected on final approach. The pilot estimated that carburettor heat was probably only on for about 10 seconds.

Approach to the 330 degree strip is over a gully. On short final the aircraft encountered some wind shear and began to sink below the glide path. The pilot pushed the throttle forward but there was no response from the engine. Further sink was encountered, and it became obvious that the aircraft was going to touch down before the airfield boundary fence. The aircraft touched down heavily, ran into the fence and slowly went over onto its back.

Post accident inspection of the engine did not reveal any mechanical reason for the lack of response to throttle application. Information from the Bureau of Meteorology showed that conditions were conducive to the formation of serious carburettor icing at any power setting. The pilot thought that because carby heat was only applied for about 10 seconds, carburettor ice was the only reasonable explanation for the loss of power.

The pilot reported that he and his 13 year old son were going for a short flight from Maryborough airport. All necessary preflight checks were carried out and sufficient fuel was on board for the intended flight. After take-off from runway 17 at between 100 and 200 ft the engine stopped completely. Witnesses saw the aircraft enter a left turn at a low height. The aircraft then appeared to stall and hit the ground inverted. The landing gear was in the retracted position.

The pilot stated that the engine failed because the magneto switches were bumped to the off position. Both switches are mounted close together in the centre of the instrument panel with other switches below them. The pilot stated that he was told in hospital that his son was waving to a person on the ground and had bumped the magneto switches off.

VH-NJG departed unrestricted on the Runway 19 Kilcoy Standard Instrument Departure (SID). The next departure off Runway 19 was VH-XFZ departing on a heading of 220 degrees. The departure restriction was to accommodate an aircraft arriving from the west for runway 14.

The Departures Controller intended to turn the trailing aircraft (VH-XFZ) further right shortly after contact was made. However, there was a Secondary Surveillance Radar (SSR) correlation problem with VH-XFZ in that its transponder did not operate until the crew recycled the selector. When VH-XFZ showed up on the SSR, it was less than 2 NM behind VH-NJG. However, the crew of VH-XFZ reported that they had VH-NJG in sight since take-off.

The Departures Controller had assumed that the ADC would provide separation of two minutes or 3 NM for departing aircraft if their tracks diverged by less than 30 degrees.

As the crew of the trailing aircraft had visual contact with the preceding aircraft throughout their take-off and initial climb, there was no risk of collision.

Analysis

Under the circumstances, the Departures Controller's operational technique did not provide separation assurance. His assumption that the ADC would apply separation was the more significant factor in the breakdown of separation.

Safety Result

Following the incident Local Operating Instructions were amended to define more clearly the responsibilities of both the ADC and Departure Controller in relation to initial departures.

On return from the training area the pilot advised that he had an unsafe indication from the nose landing gear. The tower confirmed that the nose gear was in the retracted position, so the pilot elected to return to the training area to attempt to rectify the problem. All attempts to lower the nose gear using normal and emergency systems were unsuccessful. Emergency services were placed on standby, and the aircraft subsequently landed with the nose gear retracted.

Investigation revealed that the nose gear door actuating system was worn excessively and out of rigging tolerance. This allowed the system to go to an over centre position and hold the doors in the closed position, thus preventing the nose gear from extending. The door actuating rods were incorrectly attached to the rear side of the door brackets, and the door hinges were stiff from lack of lubrication.

The pilot made a slightly heavier than normal landing. During the landing roll the nose gear down light extinguished, and the nose gear collapsed.

An investigation revealed the nose gear retraction ram mounting bracket had failed, allowing the nose gear to fold rearwards.

The aircraft had been involved in a similar occurrence 70 hours previously when the original factory installed aluminium mounting bracket failed. The bracket used to effect the repair was made from the same type of material, and had failed in a similar manner

Later models of this type of aircraft are fitted with a steel bracket.

Research of the Bureau's occurrences database, and contact with the manufacturer did not indicate this type of aluminium bracket was unduly prone to failure. Because the new stronger steel bracket is now available from the manufacturer no further safety action is considered necessary.

Approaching the Mangalore area the pilot commenced descent and had reached a height of about 1000 feet when the engine suddenly started to vibrate severely. When the throttle was closed the engine stopped completely. The pilot selected a nearby paddock and landed with a slight tailwind. During the ground roll the right wheel entered a rut, which dislodged the right main gear leg and caused the right wing and the propeller to contact the ground.

The reason for the engine vibration and failure has not been determined.

Significant Factors

The following factors were considered relevant to the development of the accident:

1. Engine power loss, reason undetermined.

2. Forced landing on unsuitable terrain.