On 18 January 1990, a Boeing 727-200 (N8867E) being operated by Eastern Airlines on a scheduled domestic passenger flight from New York La Guardia to Atlanta as EA 111 collided with a Beechcraft King Air A100 (N44UE) being operated by Epps Air on a non-revenue air taxi positioning flight from DeKalb/Peachtree to Atlanta. Both aircraft had landed in sequence off ILS approaches to runway 26R at Atlanta in normal night visibility. There was no fire and although the collision caused substantial damage to the 727, none of its 157 occupants were injured. The King Air was destroyed by the impact. Its Pilot was killed and the other occupant, a Company Pilot not rated on type and present for type familiarisation purposes only, was seriously injured.
An Investigation was carried out by the National Transportation Safety Board (USA) (NTSB). The FDR and CVR from the 727 were removed and successfully replayed but the King Air did not have either type of recorder installed and this was not a regulatory requirement. Recorded ATC radar and communications data was available and enabled the separation of the two aircraft to be seen in relation to ATC instructions and the response of the crews to them. Visibility was good and controllers in the TWR could clearly see all relevant aircraft whether on approach or on the ground.
The Investigation found that radar vectors had been issued to three inbound aircraft to establish the landing sequence for runway 26R immediately prior to the collision. The first of these was a Continental Airlines DC-9 operating under call sign CO 9687 which was followed by the King Air and the 727, which subsequently collided. At this time, simultaneous ILS approaches were underway on runways 26R and 27L and take-offs were underway on runway 26L. At the time of the accident, access to runway 26L and clearance for take-off from it were the responsibility of the same TWR controller as was handling arrivals on runway 26R - the "North Local Controller".
The ATC system in use involved inbound aircraft established on the ILS LOC being observed on radar by a controller occupying a "monitor" position in the ACC. The duties of the controller in this position were noted as being "to ensure separation between aircraft and to ensure aircraft do not enter the no transgression zone (between the runways)". The controller in this position was able to "override the tower controller and issue instructions to inbound flights".
Recorded ATC communications showed that immediately after the King Air had established initial contact with TWR, the radar monitor controller had, using the TWR frequency, asked the aircraft for its current airspeed and, on receipt of the response "175" had instructed a reduction to 160. Less than minute later, the 727 with its Captain as PF and which had, after being initially instructed to maintain 180 KIAS to the OM and then told as it approached that position only to "decrease to final approach speed", checked in with TWR. Soon afterwards, the monitor controller instructed the King Air to reduce to 150 and then 140. A landing clearance from TWR for the DC-9 was issued followed 7 seconds later by a similar (unconditional) clearance for the King Air. 42 seconds later with the 727 just 2.75 miles behind the King Air and separation slowly reducing, the TWR controller advised that it was "in sight" and cleared it to land.
It was noted that the procedures in place at the time stated that "landing clearance to a succeeding airplane in a landing sequence need not be withheld if the controller can observe the position of the airplane and determine that prescribed runway separation will exist when the airplane crosses the landing threshold" and also "requires the controller to issue traffic information to the succeeding airplane". However, no such information was issued to the 727 and FDR data showed that it had passed the OM at 165 KCAS and FDR and CVR data showed that as the Captain made a 1,000 feet callout, the speed was 149.5 KCAS and then stabilised at 145 KCAS ten seconds later, which corresponded to the calculated Vref of 134 KIAS. Once past the FAF, the King Air ground speed had progressively reduced to around 100 knots, which was considered to indicate reasonable proximity to what was calculated to have been the appropriate Vref at the aircraft's landing weight.
The evidence available showed that despite the fact that both the King Air and the 727 complied with all ATC speed control instructions, "radar data indicate that the distance between the airplanes continued to decrease" and "the last radar data about 1 mile prior to the collision point indicated 1.2 miles between the two airplanes".
As the King Air decelerated after touchdown, it was found to have moved to the right hand side of the runway as it approached taxiway 'D', the primary exit to the general aviation apron, which was about 1,160 metres beyond the 26R landing threshold. The 727 crew reported that their aircraft had touched down approximately 350 metres from the threshold after which "the spoilers were manually deployed and the nose of the airplane was lowered to the runway".Then, "as the Captain reached for the thrust reversers, another airplane was seen for the first time on the right side of the runway ahead of them […] when their landing lights illuminated it". The 727 CVR data was noted to include "two exclamations about 3 seconds prior to the sound of impact". The 727 Captain responded to the sighting by immediately steering his aircraft to the left but there was insufficient time to avoid a collision and "the right wing of the 727 struck the tail cone and structure of the King Air, separating the horizontal and vertical stabilizers from the fuselage and shearing the top of the fuselage/cockpit from the airplane". The King Air was destroyed and "came to rest at the turnoff for taxiway "D" with the engines still running". There was no post-impact fire. After the impact, the 727 Captain steered the aircraft back towards the centre of the runway to complete deceleration and the aircraft then exited the runway onto RET 'B3' and stopped. Following a transmission from the 727 that they had just hit another aircraft on the runway, the TWR controller instructed a following aircraft which had also been cleared to land on 26R to go around. The RFFS were alerted and attended both aircraft.
The 727 crew noted that the contents of one of the hydraulics was low and one of the passengers "reported a loss of some type of fluid from the right wing" and in response, the No. 3 engine was shut down followed subsequently by engines 1 and 2. The Captain decided that an emergency evacuation was not necessary and buses were used to transport passengers to the terminal.
It was found that once the DC-9 had landed, the TWR controller had attempted to pass a taxi clearance which would keep the aircraft clear of runway 26L which it would need to cross en route to the terminal in order to ensure it did not conflict with the take-off clearance he was intending to give to an Atlantic Southeast Embraer Bandeirante waiting at the threshold of runway 26L. However, he had encountered difficulties in getting this acknowledged and had become temporarily distracted just as the separation between the King Air and the 727 was continuing to reduce.
An annotated diagram showing the aerodrome layout in the vicinity of runway 26R. [Reproduced from the Official Report]
It was found that the TWR controller had only been in position for about 4 minutes when the collision occurred, having relieved the TWR Supervisor. The TWR radar monitor had been set to "a range of 18 to 20 miles" and there was a clear view of all traffic under TWR and GND control when seated.
It was found that the only external lighting being displayed by the King Air at the time of the collision had been its navigation lights. Given the prevailing runway lighting environment, it was concluded that this would have made it difficult for the 727 crew to see the King Air as it neared the right hand edge of the runway in preparation for a north-side exit to the General Aviation Apron. It was noted that it was permissible for general aviation pilots to "turn off the anti collision light system in the interest of safety if it proves distracting" and that in this respect, no distinction was made between rotating anti collision beacons and strobe lights. In the particular case of the King Air, it was noted that with wing tip strobe lights, engine nacelles and propellers all close to the pilot's position, reflections in the cloud which had probably been encountered during the inbound flight may well have been distracting and that "the pilot would most likely have turned off these lights" as a result.
No factors relating to the licensing and operational alertness of any of the pilots or controllers involved relevant to the occurrence of the collision were identified by the Investigation. It was also determined that all relevant ATC equipment had been fully operational and that the prevailing weather conditions had not contributed to the accident. However, prevailing (and generic) ATC instructions were noted as requiring that that controllers must "separate an arriving aircraft from another aircraft using the same runway by ensuring that the arriving aircraft does not cross the landing threshold until […] the other aircraft has landed and taxied off the runway".
More generally, it was noted that the Investigation had found that the collision had involved "the same controller-related factors" as those found to be causal in many previous high speed ground collisions and in events which had almost led to ground collisions. It was noted that “these factors are human performance-related and are being addressed in a number of different actions, including FAA and industry efforts to increase awareness of the magnitude and nature of the human performance problem, improved training and technological solutions that may reduce the workload, and a fail-safe redundancy for the human performance of air traffic controllers". FAA work already underway to improve timely ATC detection and response to collision risk scenarios by the deployment of new alerting systems was noted but whilst this was welcomed, it was considered that these initiatives were both long term and "intended for a limited number of high-density air carrier airports". The NTSB therefore considered that the FAA should expedite its efforts to "fund, support and implement an operational system analogous to the airborne conflict alert system to prevent runway incursion incidents at all U.S. certificated airports".
The formally-stated Conclusions of the Investigation included the following:
- The Eastern flight crew had three opportunities to learn about preceding landing traffic by listening to the tower frequencies; however, the time between transmissions, the large number of transmissions, and the required duties in the cockpit would have limited the utility of that information.
- Airspeed reduction transmissions to N44UE by the radar monitor controller were insufficient to achieve the required 4 miles separation from the preceding airplane, CO 9687, on the final approach and at the threshold.
- The absence of appropriate airspeed reduction instructions to EA 111 by the Atlanta approach north final and the radar monitor controllers led to a speed differential that resulted in a loss of the separation between EA 111 and N44UE.
- The traffic volume at the time of the accident presented an average controller workload, but the local controller was distracted with radio difficulties (misunderstood instructions) when communicating with CO 9687.
- The local controller's distraction by communication difficulties with CO 9687 was prompted by his perceived need to clear runway 26 left for another airplane inbound with a hydraulic emergency and the possibility of a runway incursion from CO 9687 during taxiing. The controller was inattentive to the more immediate task of monitoring the separation of traffic landing on runway 26 right.
It was determined that the Probable Causes of the Accident were:
- the failure of the Federal Aviation Administration to provide air traffic control procedures that adequately take into consideration human performance factors such as those which resulted in the failure of the north local (TWR) controller to detect the developing conflict between N44UE and EA 111.
- the failure of the North Local (TWR) Controller to ensure the separation of arriving aircraft which were using the same runway.
Two Contributory Factors were also identified:
- the failure of the North Local (TWR) Controller to follow the prescribed procedure of issuing appropriate traffic information to EA 111.
- failure of the North Final (APP) Controller and the Radar Monitor Controller to issue timely speed reductions to maintain adequate separation between aircraft on final approach.
Five Safety Recommendations were made at the conclusion of the Investigation as follows:
- that the Federal Aviation Administration should develop an Air Traffic Bulletin and provide a mandatory formal briefing to all air traffic controllers on the importance of and the need for giving air traffic information when issuing an anticipated separation and landing clearance. [A-91-27]
- that the Federal Aviation Administration should amend the Air Traffic Control Handbook, 7110.65F paragraph 3-127, to preclude the issuance of multiple landing clearances to aircraft outside of the final approach fix. Also establish a numerical limit so that no more than two landing clearances may be issued to successive arrivals. [A-91-28]
- that the Federal Aviation Administration should expedite efforts to fund the development and implementation of an operational system analogous to the airborne conflict alert system to alert controllers to pending runway incursions at all terminal facilities that are scheduled to receive Airport Surface Detection Equipment (ASDE-3). [A-91-29]
- that the Federal Aviation Administration should conduct research and development efforts to provide airports that are not scheduled to receive Airport Surface Detection Equipment with an alternate, cost effective, system to bring controller and pilot attention to pending runway incursions in time to prevent ground collisions. [A-91-30]
- that the Federal Aviation Administration should incorporate into the training syllabus at the Federal Aviation Administration's Academy at Oklahoma City, Oklahoma, the importance of and the need for giving traffic information when issuing an anticipated separation landing clearance. Stress that this information will enhance pilot awareness and visual acquisition of preceding traffic, thereby providing a redundancy in separation assurance for controllers and pilots. [A-91-31]
The Final Report of the Investigation was adopted by the Board on 29 May 1991 and subsequently published.