The captain did not have a track record of having problems dealing with engine failures during training but this incident was his second where he appeared unable to cope when it happened in an airplane. The first officer appeared to be doing a competent job of flying the airplane during the engine failure but completely disengaged once the captain took over. This engine failure could have been survived had the captain applied proper stick and rudder inputs and simply left the failed engine alone. Why do some pilots with adequate simulator training records do so poorly in aircraft? I think it may be a form of panic.

— James Albright





Airlink 8911 Wreckage,
from SACAA Report

A transport category airplane certified under 14 CFR 25 is designed to fly with the critical engine out, so if anything kills you, it is probably you. The two most likely ways for that to happen is when you (a) apply the wrong rudder, or (b) shut down the wrong engine. Too many pilots think of an engine failure as a macho test of speed to see who can get the rudder in so fast that other pilots don't even notice the yaw. If the airplane doesn't have time to yaw, you aren't doing it right. Now all of my engine failure experience has been in two or four-engine jets. You should practice this in a simulator, but this has always worked for me:

  1. Allow the yaw to develop, this gives you time to think and will provide a needed clue as to what is going on.
  2. Level the wings with ailerons, and unless climb performance is critical, do not rush in with rudder.
  3. Note which hand is slow (if you have a yoke) or which way the hand is pointed (if you have a stick). This is where your foot will need to be to correct the yaw. "Step on the low hand" as well as "step on the ball" of your slip indicator. Do not stomp on the rudder, you could lose the tail. Your application should take at least a second, two would be better.
  4. Wait until you are at least 1,500' before shutting down engines.
  5. Make an engine shutdown a CRM process that includes a "wait and see" step:
    1. Put your hand on the throttle you intend to pull and ask the other pilot to confirm, "I suspect number one is failed, I have my hand on the number one throttle. Confirm?"
    2. Pull the throttle to idle and wait. Verify the situation didn't get worse.
    3. Announce the throttle pull reaction and your intention to shut down the engine, get validation. "I've pulled number one to idle, the yaw hasn't changed and we are still climbing. I am about to shut down number one. Confirm?

1 — Accident report

2 — Narrative

3 — Analysis

4 — Cause



Accident report

  • Date: 24 September 2009
  • Time: 07:57
  • Type: British Aerospace 4121 Jetstream 41
  • Operator: SA Airlink
  • Registration: ZS-NRM
  • Fatalities: 1 of 3 crew, 0 of 0 passengers
  • Aircraft Fate: Destroyed
  • Phase: Takeoff
  • Departure Airport: Durban-Louis Botha Airport (DUR/FADN), South Africa
  • Destination Airport: Pietermaritzburg Airport (PZB/FAPM), South Africa



  • The recorders provided confirmation that the aircraft was configured for a flap-9 takeoff and that initially the copilot was the pilot flying. The initial takeoff roll appeared normal, but as the aircraft accelerated through about 90 kt, 5 kt below V1, the right-hand engine torque started to reduce. As it dropped, the associated engine RPM remained at about 100% and there were no recorded warnings generated at that time. Evidence from the CVR indicates that it was at about this point that a transmission from another aircraft was made on the tower frequency, advising that smoke was emanating from ZS-NRM. The ATC relayed this information to ZS-NRM, but the commander simultaneously called "V1, rotate" as the aircraft accelerated through about 95 kt.
  • The aircraft became airborne at about 125 kt. Seconds later, as it was climbing through a height of about 100 ft above mean sea level (AMSL), there followed the first of a series of flight deck aural attention chimes. The first of these was confirmed by the copilot as being due to "right oil contamination". The aircraft continued to climb and about five seconds after the copilot's comment, the captain stated: "We have lost an engine, we are losing an engine". The copilot responded: "I have it, I have it, keeping runway track 6 000 ft. Flap is zero. We have lost an engine."
  • The aircraft continued to climb, but as the right engine torque reduced below 20%, the airspeed started to decay; the maximum airspeed recorded being about 145 kt at 185 ft AMSL. As the aircraft approached about 400 ft AMSL, the right engine torque had reduced to 0% and the airspeed was reducing through 132 kt. This was followed by a gradual reduction in right-engine RPM. At 440 ft AMSL, the flaps were retracted, by which time the aircraft had begun to roll progressively and turn to the right despite both left rudder and left aileron being applied.
  • At 490 ft AMSL, the aircraft momentarily levelled out, with the airspeed now reducing through about 120 knots. At about this point, the copilot stated: "We're not maintaining", which was acknowledged by the captain. This was followed by the sound of the master warning activating. At the same time, the right engine Beta discrete value indicated zero. An unidentified radio transmission also advised: "Your gear is still down." The captain was then again heard to say: "OK", just before the left engine torque and RPM indications rapidly reduced to 0%, accompanied by left- engine low oil pressure and hydraulic low pressure warnings – consistent with the left engine having been manually shut down.
  • The copilot could be heard calling for the gear to be raised, which the captain acknowledged. Further alerts could also be heard sounding. The aircraft had started to descend and as the angle of attack, which had been gradually increasing, reached approximately 14°, the stick shaker activated.
  • At this point, the copilot referred to the captain by name, saying: "Pitch forward." There was no recorded handover of control, although it appears that from this point the captain was the handling pilot. There was also no recorded acknowledgement following the taking over of control by the captain.
  • The aircraft continued to descend and on passing 400 ft AMSL, the right-engine low oil pressure warning activated. Various ground proximity warnings could be heard on the CVR, together with occasional stick shaker activations, until the aircraft struck the ground. The FDR stopped recording approximately two seconds before impact due to its power supply being lost as the engines ran down.

Source: SACAA Report, ¶1.11.4



  • After rotation, the failure of an engine was identified. However, key points in the appropriate SOPs were not followed. The copilot correctly identified the oil contamination warning referring to the right-hand engine, and it should have been apparent that the aircraft was rolling to the right as well. This, together with the different torque indications, should have indicated which engine had failed, yet there appears to have been no positive attempt between the pilots to identify the problem. There was a gradual loss of control with the captain seemingly attempting to feather the propeller on the failed engine quickly in order to regain control. In doing so, he inadvertently selected the wrong engine. Having done so, a forced landing became inevitable.
  • The actions of the pilots were possibly affected by the reports of smoke and the gradual loss of control, both of which would have created a heightened degree of urgency. The operator's training procedures were apparently adequate and neither pilot had any major problems identified in their training records. Both pilots should have been capable of successfully handling the failure.
  • The captain had been involved in a very similar accident on 21 August 2005. The ATC had cleared the captain to land on runway 05 at Virginia Aerodrome in KwaZulu-Natal Province. When the captain was on short finals, the tower noticed that the aircraft was drifting away from the runway centreline and called the pilot. He stated that he was experiencing an engine problem and was initiating a go-around. The aircraft turned out to the left and away from the runway centreline, and the pilot allowed the aircraft to continue flying over the nearby M4 motorway and then towards a residential area. The aircraft struck the roof of a private home and came to rest in a tail-high, inverted position.
  • It thus appears that on both occasions the pilot had difficulty in controlling the aircraft after an engine failure.
  • The No. 2 or right-hand engine failed on rotation and a power reduction occurred on the No. 1 engine as the aircraft climbed to about 480 ft AMSL. The aircraft was seen to climb and thereafter descended and struck the ground. The total time from start of the take-off roll until impact was about 110 seconds.
  • Analysis of the FDR data (Figure 4) revealed that during take-off, at about the moment of rotation, five seconds before actually lifting off the runway, the right-hand engine failed and the thrust gradually reduced to zero in about 25 seconds. The take-off was continued. As shown by the FDR data, the dynamic effects caused by the failing engine were minimal, although the aircraft entered two small rudder pulses to the right immediately after the engine failure. The aircraft started climbing and the captain initially maintained approximate runway heading with a small, but increasing rudder input to the left. The bank angle (roll) varied slightly around wings-level; roll control power by the ailerons was adequately available. Despite the reduction of thrust from the right-hand engine, the airspeed continued to increase for about 12 seconds to approximately 135 kt, which was still much higher than the required V2 (103 kt) for this flight phase. The captain increased the rudder deflection for maintaining the heading, but allowed a small bank angle to the right, into the inoperative engine, rather than away from it. Consequently, the drag must have increased and as a result, the airspeed started to decrease. The bank angle could still be controlled using the ailerons, but was not five degrees (or less, as is preferred by the manufacturer) away from the inoperative engine, as is required for minimum drag and lowest possible minimum control speed while the thrust is asymmetrical.
  • When the decreasing airspeed reached approximately 127 kt (at 05:57:22), the aircraft started rolling to the right. Instead of deflecting the ailerons to the left in the first place as required to prevent the roll, the pilot had allowed the roll to begin. Left aileron input was then increased to 10° out of a maximum 14°, but this was insufficient, at the decreasing actual airspeed, for the ailerons to generate a high enough rolling moment to counter the propulsive thrust rolling moment generated by the blown wing section behind propeller No. 1. Lateral (roll) control was lost at this point (05:57:22). The indicated airspeed had decreased below the actual lateral minimum control speed for the given (less than maximum) aileron deflection and thrust setting.
  • The torque of the No. 1 (left) engine, which was slightly reduced to approximately 85% for unknown reasons, was then slowly increased to over 100% in 10 seconds. The increased engine torque increased the thrust yawing moment (and the rolling moment due to thrust) even more. An increase of opposite rudder deflection can be observed, but was not high enough (it was not more than 12° of the available 24°) to prevent the yawing from increasing to the right and maintain the heading at and below airspeed of about 125 kt. Directional control was lost as well at this point (05:57:23). The actual (directional) minimum control speed with only 14° of the available 24 degrees of rudder deflection, the actual engine thrust and the actual bank angle was 125 kt.
  • Just after the torque increase started (at 05:57:23), the flaps were selected up. The flaps may have an influence on VMCA (minimum control speed) because the effect of the airflow striking the vertical tail. The flap handle might also be mechanised to switch on or increase the rudder boosting; this will have an effect on the value of VMCA and thereby the controllability of the aircraft.
  • From 05:57:23, the aircraft kept rolling and yawing to the right, despite opposite control inputs. The forces generated by these limited (less than maximum) control deflections and the actual airspeed were obviously not high enough to counter the propulsive thrust rolling and yawing moments. Since both lateral and directional control were lost at an indicated airspeed of about 125 kt, this airspeed was obviously the actual VMCA of the aircraft at that instant, with the actual values of power setting, control deflections, bank angle, centre of gravity, weight, etc. This VMCA was higher than that listed in the aircraft flight manual because the actual bank angle and control deflections were not the same as those used to determine the VMCA and take-off speeds (VR and V2) in the manual. In addition, VMCA could have been increased by flap retraction (at 05:57:23) or by gear retraction (at approximately 05:57:30), thereby lessening the controllability of the aircraft.
  • At 05:57:31, the torque of the other engine also suddenly decreased to zero within ten seconds. Following this total power loss, there was no asymmetrical thrust anymore, and hence no adverse thrust yawing and rolling moments.

Source: SACAA Report, ¶2



  • Engine failure after takeoff followed by inappropriate crew response, resulting in the loss of both lateral and directional control, the misidentification of the failed engine, and subsequent shutdown of the remaining serviceable engine.
  • Failure of the captain and first officer to implement any crew resource management procedures as prescribed in the operator's training manual;
  • The crew's failure to follow the correct after take-off engine failure procedures as prescribed in the aircraft's flight manual.

Source: SACAA Report, ¶3.2


(Source material)

South African Civil Aviation Authority (SACAA) Accident Investigation Report CA18/2/3/8690, Jetstream Aircraft 4100 ZS-NRM: Loss of control after engine failure and misidentified engine shutdown after take-off from Durban Airport, South Africa on 24 September 2009