- Luxair Flight 9642 at Luxembourg
- Accident Overview
- Accident Board Findings
- Accident Board Recommendations
- Relevant Regulations / Policy / Background
- Prevailing Cultural / Organizational Factors
- Key Safety Issue(s)
- Safety Assumptions
- Resulting Safety Initiatives
- Airworthiness Directives (ADs) Issued
- Common Themes
- Related Accidents / Incidents
- Lessons Learned
- Luxair Flight 9642 at Luxembourg
Photo of accident airplane in Paris, October 18, 2002
History of Flight
A Fokker F27 Mk 050, registered as LX-LGB and operated by Luxair, left Berlin for Luxembourg on Flight LG 9642/LH 2420 with 19 passengers and three crew members (captain, first officer and flight attendant) on board. Initial cruising altitude was flight level (FL) 180 - 18,000 feet. While en route the crew checked the destination weather, and learned that the visibility was 100 meters with a runway visual range (RVR) of 250 meters in fog. The flight crew discussed the possibility of delays, and whether they would enter a holding pattern at the destination, or divert to an alternate airport. No decision regarding the approach was made at the time, but no approach briefing was conducted in preparation for an approach.
Photo of Fokker F27 Mk 050 cockpit
Photo copyright Jonathan Rankin - used with permission
Accident Flight Path
(View Larger Image)
On initial contact with Luxembourg approach, the flight was instructed to enter the FL90 (9,000 feet) holding pattern at Diekirch VOR, and to expect radar vectors later for an instrument landing system (ILS) Category II approach to Runway 24 (A 4,000 meter long runway at an elevation of 1,214 feet and oriented on a heading of 241°).
During the initial stages of the approach, the pilot and co-pilot discussed concerns regarding the expectation that they would be instructed to hold and the amount of fuel they had available before a diversion to an alternate airport would be necessary. As the flight neared the point where it would enter the holding pattern, the controller cleared the flight to initiate an approach rather than enter the hold. The investigation concluded that this was an attempt by the controller to maintain separation between all aircraft in the hold at the time, and alleviated the need to assign a new and separate holding altitude for the accident flight.
The flight crew was instructed to contact the tower approximately simultaneously with interception of the ILS beam (localizer and glideslope). The crew contacted the tower and reported that they were established on the localizer and were continuing the approach. Following this initial contact, a discussion between the tower and flight centered on the necessity for a visibility (RVR) of 300 meters to continue the approach and land.
RVR is measured from an automated instrument known as a transmissometer, which is located to the side of the runway. The transmissometer measures the maximum distance (in feet or meters) at which the runway lights or surface markings can be seen before they are lost in the fog or rain. Thus, it will be certain that on landing a pilot's visual range will be at least that reported distance (horizontal visibility) as the airplane rolls down the runway.
Some airports have only one transmissometer, placed near the touchdown (or initial) area of the runway. There might be another located at the midpoint (intermediate) section of the runway and/or at the far end (stopping end or rollout area). The tower issues the current RVR reading(s) and the pilot must determine if that visibility is at least the value required by his approach chart. If so, he may continue the approach and land. For example, if the pilot's approach chart requires 300 meters (1,000 feet) RVR, and the tower says the RVR is 250 meters (800 feet), then the flight crew must either wait for the visibility to improve or fly to their alternate airport. However, even if the RVR is measuring the legal visibility to begin the approach, when the airplane reaches the minimum descent altitude the pilot must still acquire visual contact with the approach lights in order to land. If the pilot does not have visual contact with the lights, a climb (go-around) must be initiated for the purpose of either executing another approach or fly to the alternate airport.
At 09:03:18, approximately two minutes before the crash, the tower controller transmitted: "9642 copied... uh, so continue approach and I'll keep you advised we didn't have 300 uh... uh during the last time."
At 09:04:36, The tower controller informed another flight, "We have now on the three positions 275 meters." This message was heard by the crew of Luxair 9642 as they passed overhead ELU, maintaining 3,000ft. ELU was a decision point by which the flight should have either intercepted the glideslope or initiated a go-around.
At 09:04:46, the captain then announced to the copilot: "Yes, well we do a go-around, missed approach."
At this point, investigators concluded that the captain initiated a missed approach by advancing the power levers.
At 09:04:57, the tower controller transmitted an RVR to the crew, "Luxair 9642 RVR 300 meters 275 meters stop end 275 meters."
Photo of aisle stand of Luxair Fokker 50
Photo copyright Brunelle David - used with permission
At 09:05:00 the Cockpit Voice Recorder (CVR) sounds indicated a variation in the rotation speed of the turbines, and the primary lock (i.e. ground range selector, which allows the power levers to be moved into the reverse pitch range) was lifted/removed. The Flight Data Recorder (FDR) indicated the start of engine power reduction.
The RVR value provided at 09:04:57, about 11 seconds after the captain said he would perform a "missed approach," corresponded exactly to the required landing minima. Investigators believed that this information triggered a sudden reversal of the captain's decision, and he resumed the approach without announcing that he was doing so. No procedure existed to capture the glide slope from above after having passed the final approach point.
The airplane performance aspects of a long landing are described in the following animation: (Landing Performance Animation)
The captain brought the power levers to flight idle, and at the same time pulled the ground range selectors in order to be able to bring the power levers slightly further backwards. This action was deduced by the investigators, based on the values of the left and right high pressure turbine RPM parameters, which were below the flight idle minimum, and by the identified relevant noises on the CVR. The "secondary stop" installed on the engines maintained the power levers in a position slightly below the flight idle. The investigation concluded that the pilot's actions to reduce the power levers to a slightly lower position than the flight idle position was an attempt to increase the descent rate without accelerating, while intercepting the glide slope. Post accident investigation showed that it actually did not improve the deceleration.
At 09:05:05 the copilot responded to the tower controller, "9642 Roger so we continue," indicating that they would be continuing the approach.
At 09:05:08 the tower controller cleared Flight 9642 for landing.
At 09:05:13 the co-pilot acknowledged the landing clearance. This was the last communication with ATC.
Flight deck view of Fokker 50 on approach
Photo copyright Serge Bailleui - used with permission
At 09:05:16 the FDR indicated the extension of the landing gear. At this point the FDR also indicated a heading of 238°, an indicated airspeed of 145 knots, and an altitude of 2,635 feet. Propeller torques (left and right) were 0% and 0 %. Propeller speeds (left and right) were 85% and 85%.
At 09:05:17 CVR sounds indicated an increase of rotational speed of at least one propeller and then numerous noises of selections, and power variations were recorded. The FDR indicated that the left propeller "blade angle" parameter switched from "normal" to "low pitch," signaling a propeller blade angle setting of less than 10°. The investigation concluded that the change in blade pitch from normal to low pitch was the result of the removal of the secondary blade low pitch stop, which had previously prevented the blade angle transit, even though the power levers were in a range where low pitch was commanded.
Post accident investigation demonstrated that the most probable cause for the removal of the secondary stop was the extension of the landing gear, which triggered the energizing of the flight idle stop solenoid relay due to a known design flaw in the antiskid control box. The investigation concluded that with all the mechanical locks having been removed, the captain continued to apply hand pressure on the power levers, and may have unintentionally moved them further backwards without realizing that he was now in beta mode, having passed through the ground idle position towards full reverse.
Intended vs. Actual Flight Path
(View Larger Image)
At 09:05:18 the FDR indicated the right hand propeller "blade angle" parameter switched from "normal" to "low pitch."
At 09:05:19 the captain says, "What's that?" CVR sounds indicated throttle going to, or through, the ground idle position. Reverse power was applied to both engines, which was documented by a rapid increase of all engine parameters. The power levers were then set beyond the flight idle position back into the flight range.
At 09:05:20 the FDR indicated the start of flap retraction. Two seconds later, both propeller speeds had increased, and the left engine was shut down.
At 09:05:23 the left propeller RPM drops below 50% and the left engine generator is taken offline by the generator control unit (GCU). As a result, the right engine generator subsequently powered all electrical buses.
At 09:05:25 the left engine high pressure rotor speed drops below 60%. No alert chime was recorded on the CVR, confirming that the engine was shut down manually. The right propeller speed had reached 108% RPM, representing the maximum value allowed by the propeller overspeed governor. At this time, the right engine was also shut down manually.
At 09:05:26 the FDR stopped recording.
At 09:05:27 the CVR captured the beginning of a ground proximity warning system alarm. One second later the CVR stopped recording. Altitude at the cessation of recording was 2,000 feet.
Photo of cockpit area at crash site
The events following gear extension at 09:05:16 happened in a very rapid sequence. The investigators believed that the increase in reverse power triggered a propeller overspeed that was heard and noticed by the crew. Feeling an increase in drag and the consequent deceleration, one of the crewmembers retracted the flaps. The power levers were moved back into the flight range but the right propeller remained in the beta range. The left engine had been shut down, followed a couple of seconds later by the right engine. The FDR and CVR readings stopped at this moment. Due to the lack of data (due to loss of electrical power to the recorders), it was not possible to analyze the subsequent flight phase. The investigators concluded that the aircraft descended without power into the fog layer, and the crew may have attempted to flare the aircraft at the last moment when they saw the ground, just prior to the crash.
Wreckage of the flight was immediately found in a field 700 meters to the north of the runway 24 extended centerline and 3.5 kilometers to the east of the threshold.
Photo of the crash site
The aircraft touched down on a heading of approximately 295°, as indicated by the general direction of the debris. The first impact marks were found on the south edge of the road RN1. They represented the two main landing gears and the fuselage tail cone. Scraping marks on the road, notably from the left wing tip, showed that the aircraft slid across the road before hitting an embankment at the north side of the road RN1. The major part of the damage results from this impact during which the aircraft lost three blades from the right propeller and two from the left propeller as well as wheels from the left and right landing gear. Furthermore, the aft portion of the fuselage was disrupted at the trailing edge of the wings by this shock. After this bounce, the empennage and part of the right outboard wing broke away, the aft portion of the fuselage turned around to the right, and the aircraft came to rest 25 meters further away in a field.
An animation of the final minutes of the flight path is available here: (Flight Path Animation)
Views of crash site
Photo of Pratt & Whitney PW125 engine and dowty
R352/6 propeller - Propeller shown in feathered position
Photo copyright M. Radzi Desa - used with permission
Power Levers and Ground Range Selector
(View Large Image)
Flight Idle Stop Solenoid
(One per Engine - General, and Close-up Views)
(View Large Images)
Diagram of Propeller Control Logic
(View Large Image)
Propeller Control System
The PW 125B engines each drive a variable-pitch, constant-speed, six-bladed propeller. A propeller speed tachometer is located on the center main instrument panel. There are two control modes for the propeller:
- Above flight idle, constant speed control is regulated automatically in flight.
- On the ground, below flight idle and in the beta mode (reverse pitch) range, propeller pitch is directly controlled by the power lever position.
The actual blade angle at which the propeller produces zero thrust (torque) depends on the rotational speed of the propeller and the airplane forward speed. At a blade angle of approximately 26°, the propulsive force (thrust vector) acting on the airplane tends towards zero, and begins reversing direction if propeller pitch is further reduced.
In flight, power lever positions below flight idle are prevented by two means:
- A mechanical primary stop (ground range selector) on the power levers. This primary stop requires a positive, distinct and separate pilot action.
- An electrical secondary stop (flight idle stop solenoid) on each engine.
To select the beta mode (reverse pitch) after landing, with the power levers in the flight idle position, the pilot lifts the ground idle selectors and moves the power levers aft. The primary mechanical stop on the power levers is supplemented by a flight idle stop fixed to each engine and activated by solenoids. Once the solenoids are energized, the flight idle stops are moved, and power levers may be moved aft, into the range for reverse thrust. Power to the solenoids is supplied when:
- One of the sensors mounted on the shock absorbers of the left and right main landing gear detects a compression of the shock absorber during landing, or;
- The two wheel speed sensors, each one mounted in the wheel axle on one main landing gear, detect a wheel speed in excess of 17 knots.
Above flight idle, the Propeller Electronic Control (PEC) unit controls propeller speed by varying the blade angle. Speed is controlled to 100% during take-off, maximum continuous, and go-around power settings. Propeller speed is controlled to 85% during climb and cruise. Propeller synchronizing is totally automatic.
Propeller pitch angle varies in flight from + 15° to approximately + 45°. Propeller pitch is controlled by balancing oil pressure against the coarse "seeking force" that results from the counterweights, which are attached to the roots of the blades. A high-pressure pump, driven by the propeller gear box and supplied with engine oil, provides the required oil pressure. In the event of an oil pressure loss, the counterweights will move the blades to an angle of + 55°, thus preventing propeller overspeed and minimizing the drag created by the windmilling propeller. The dedicated drive of the high-pressure pump assures control as long as the propeller is windmilling. In case of an in-flight engine failure, the propeller control mechanism initially tries to maintain a constant speed of the propeller in relation to the airspeed until it is feathered, either automatically or manually.
Photo of F27 Mk 050 aisle stand showing
power levers and ground range selectors
Power Lever Angle vs Propeller Blade Angle
(View Larger Image)
Below flight idle, the power lever controls propeller pitch directly from a blade angle of approximately +15° to -17° (full reverse). In the beta mode, the commands of the propeller electronic controls are inhibited. Propeller blade angles are then solely controlled by the movement of the power levers (power lever angle). A blue low pitch light, located on the central instrument panel, comes on when the blade angle drops below 10°. During the investigation it was found that, while in flight with the power levers in flight idle, it is possible to lift the ground range selectors (primary lock) and move the power levers a small distance further aft until blocked by the secondary lock. This requires a deliberate action by the pilot, and is prohibited by the flight manual.
In flight a propeller overspeed governor is activated when propeller speed reaches 104%. The gearbox-driven governor reduces the oil flow to the pitch changing mechanism. If there is no propeller speed reduction, the propeller speed reaches 108% and the overspeed governor intervenes directly in reducing fuel flow. On the ground, with the propeller in reverse pitch, overspeed protection is accomplished at 108% by reducing the fuel flow.
The propeller can be feathered either automatically or manually. The propeller is feathered manually when the fuel lever is set to SHUT or START. The feathering pump is activated when:
• The autofeathering system is activated when the aircraft is on the ground or in flight, or;
• When the fuel lever is set to SHUT or START when the aircraft is in the air. The feathering pump brings the blade pitch angle to a position of 82° in order to minimize aerodynamic drag (feathered propeller).
Photo showing right engine/propeller
propeller in reverse pitch
A brief and very basic tutorial on relevant propeller control concepts and terms is available at the following link: Propeller Tutorial
An animation of the operation of the propeller in this accident is available at the following link: Propeller Animation.
Electromagnetic Interference (EMI)
Electromagnetic interference is the disruption of operation of an electronic device when it is in the vicinity of an electromagnetic field (EM field) in the radio frequency (RF) spectrum. EMI can affect an electrical circuit due to either electromagnetic conduction or electromagnetic radiation emitted from an external source. The disturbance may interrupt, obstruct, or otherwise degrade or limit the effective performance of the circuit. The source may be any object, artificial or natural, that carries rapidly changing electrical currents, such as an electrical circuit, the sun, or the Northern Lights. EMI can occur unintentionally as a result of spurious emissions. It frequently affects the reception of AM radio in urban areas, and can also affect cell phones, FM radio, and television reception.
Radiated EMI may be broadly categorized into two types; narrowband and broadband.
Diagram of Electromagnetic Interference – Radiated and Conducted Sources
Narrowband interference usually arises from intentional transmissions such as radio and TV stations, pager transmitters, cell phones, etc. Broadband interference usually comes from incidental radio frequency emitters, such as electric power transmission lines, electric motors, thermostats, etc. Anywhere electrical power is being turned off and on rapidly is a potential source. Moderate- or high-powered wireless transmitters can produce EM fields strong enough to upset the operation of electronic equipment in close proximity. Cordless telephones, home entertainment systems, computers, and certain medical devices can fail to work properly in the presence of strong EM fields.
Conducted Electromagnetic Interference is caused by the physical contact of the conductors, as opposed to radiated EMI, which is caused by induction (without physical contact of the conductors). Electromagnetic disturbances in the EM field of a conductor will no longer be confined to the surface of the conductor and will radiate away from it. This persists in all conductors and mutual inductance between two radiated electromagnetic fields will result in EMI.
Existing Propeller Control System Anomaly
An EMI problem with the secondary flight idle stop on the F27 Mk 050 was identified as early as 1988. As the antiskid units powered up in response to landing gear extension, it was learned that if both units powered up within 20 milliseconds of each other, EMI resulting from this power up sequence could cause the ground control relay to activate, in turn causing the secondary lock to briefly release. In this situation, the secondary lock would be released for approximately 16 seconds. This could therefore result in a hazardous situation if the primary stops were intentionally released, and the power levers were held against the secondary stop with some force. This action was strictly prohibited by flight manual procedure, as follows:
Photo of accident airplane on approach
Photo copyright Sven De Bevere - used with permission
"Do not attempt to select Ground Idle in flight. In case of failure of the flight idle stop, this would lead to loss of control from which recovery may not be possible."
When the service bulletin detailing modifications to correct inadvertent EMI-induced secondary stop activation was first introduced in 1992, it was not mandated. In 1993 Fokker received an incident report of power lever selection below flight idle during approach. Also, several airlines reported that ground range selector levers (primary stops) had been operated occasionally during flight, primarily in turbulence. In 1994 Fokker informed operators about the possibility of EMI-induced in-flight release of the beta lockouts, and recommended incorporation of the modification when an antiskid unit was removed or repaired for another reason. Following the accident, the modification was mandated by the European authority.