Atlantic Airways Flight 670

Investigation

The investigation was carried out by the Accident Investigation Board Norway (AIBN), who arrived by helicopter at the scene at 13:08. Three individuals filmed the accident, of which one was especially useful: from 13 seconds after the aircraft left the runway and subsequent 21 minutes was filmed by a witness across Stokksundet, 1.5 km from the site of impact. He sold the tape to TV 2, which subsequently handed the video over to AIBN. The investigators concluded that the runway was damp when they arrived, but were not able to conclude if it was wet at the time of the accident. It was nearly impossible to carry out a meaningful investigation of the fuselage because of the heat damage, although the detached left landing gear could be investigated.

The commission mapped all skid marks on the runway. The first skid marks found to be by OY-CRG was at 945 m past the threshold for runway 32. The aircraft also left rubber tire debris. Initially following the centerline, the aircraft was found to drift to the right after 1140 m and then shifted direction after 1206 m. From 1274 m the aircraft was skidding to the left, gradually reaching an angle of twenty-five degrees, at the time it ran off the runway at 1465 m.

The flight data recorder was recovered but had sustained substantial fire damage, exceeding its design limits. The tape-based Plessey Avionics PV1584J was sent to the Air Accidents Investigation Branch for investigation. They were only able to extract three segments of the flight, one hour during the flight from the Faroe Islands to Stavanger; 12 seconds during the approach to Sørstokken, ending 43 seconds before the final segment, which lasts for 3 seconds, ending three seconds before the end of the recording. The Fairchild A100S cockpit voice recorder used a solid-state storage medium. It was sent to the same laboratory, but no data could be retrieved there due to fire damage in the circuit board. However, when sent to the manufacturer, they were able to apply repairs to successfully retrieve the content. The sound files were sent to the Accident Investigation Board Finland, who were able to establish a timeline and that the spoiler lever had been set correctly.

All cockpit communication was strictly related to the flight and proper crew resource management was executed. The pilot stated that he believed the aircraft had sufficiently low speed that it would have stopped had the runway been 50 to 100 m longer. The first officer estimated the speed at the overrun time at 5 to 10 km/h and that the aircraft would have stopped had the runway been 10 to 15 m longer.

The six spoiler actuators were sent to the Royal Norwegian Air Force's facility at Kjeller for examination. Radiographic examinations confirmed that they were all in a closed and locked position. A simulation was carried out in a flight simulator to see if, given the conditions, a 146 could land at Sørstokken without operative spoilers. It concluded that this would be possible with a dry runway, but not if it is wet. An extensive investigation of the spoiler system was carried out by Aviation Engineering, who published their findings on 10 May 2011. AIBN quickly started working out from a hypothesis that the spoilers did not deploy and investigated three possible causes: a mechanical failure in the lever, a failure in two thrust lever micro switches, and an open circuit breaker in the lift spoiler system. In the latter case, it would have been necessary for two of four to fail although one could have failed without being detected.

The commission found that the approach and landing were normal with natural variations but that the spoilers were not deployed when the pilot pulled the lever. No conclusion to the cause was found, although the commission believed it must either have been a mechanical fault in the spoiler lever mechanism or a fault in two of the four thrust lever micro switches. The pilots received warning of the failure of deployment and also noticed the lack of sufficient retardation, but failed to connect the two issues, instead focusing on the wheel brakes. The pilots perceived they would not stop in time and activated the emergency braking system. This system bypasses the antilock braking system and can lock the wheels completely. In this incident the wheels locked and the tires rapidly heated due to friction with the runway surface. Combined with damp conditions on a non-grooved runway, this situation resulted in reverted rubber hydroplaning, a state where the heated rubber created a layer of steam between the tires and the runway thereby further significantly decreasing the effectiveness of the braking system and adding approximately 60% to the distance required to stop. The lack of runway grooving was decisive for the hydroplaning to take place. The aircraft was estimated to be traveling at 15 to 20 kn at the time of the overrun. Had optimum braking taken place, the aircraft would probably have come to a stop on the runway. The massive damage was not caused by the overrun as such, but rather the steep slopes on the side of the runway.

Further, the commission found that the fire was caused by a fuel leak ignited by a short circuit. This spread to the fuel tanks, leading to the scale of the inferno. Supply of ample oxygen was provided by an engine still running. All people on board had a chance of survival, but that rapid and correct evacuation was decisive for actual survival. Although the fire crew were quick to the scene, they were hindered by the terrain, preventing efficient containment of the fire during the crucial evacuation period. The aircraft and crew were found to be fit, airworthy and certified. No training or procedures were available in case spoilers failed to deploy, and that these could have prevented the accident. The airport's physical geography and lack of adequate safety were decisive in the outcome of the accident.