Tuesday, September 23, 2008
Sunday, September 14, 2008
It must be understood that both the Captain and the F/O are human beings and that their performance is susceptible to all human limitations. These include being prone to loss of situational awareness, stress and fatigue; having an attitude and perceptual problems. Other problems may include limitations in the application of knowledge, exercise of judgment, short term memory (working memory, which is akin to the RAM in a computer) overload, and finally incapacitation. Also, we are all familiar with the saying that ‘to err is human’. This saying will continue to be true as long as we human beings have limitations. However, in aviation – a profession that is very constrained by time; is highly stressful; has complex processes; and which is closely scrutinized the media, regulator and public, we can ill afford to have human errors, as these errors can be fatal, and have been so in the past. Human errors are defined as ‘the unintentional act of performing a task incorrectly which can potentially degrade the system’. Errors can include problems in practice, procedures, and systems. Human error has been a major contributor to incidents/ accidents in the past. Here it also needs to be emphasized that accidents do not generally happen due to a single factor. They are a culmination of a ‘chain of events’ of small factors. The aircrew is the last link in this chain that can prevent these errors from turning into incidents/ accidents. To prevent these incidents/ accidents, the crews need to be aware of error management.
Error management can be best accomplished by ‘Error Avoidance’. In case this is not possible then the next step would be ‘Early detection’. Finally, in case the error has been detected late or not detected, then steps have to be taken to ‘minimise the consequences’ as a result of these residual errors. Let’s take an example to understand this process. We have been cleared for a radar vectored ILS approach for R/W 28 at Delhi. The local QNH is 983 hpa and the weather is Rain/ Thundershowers with low clouds at 300ft. We are descending through transition level, and about to change the altimeter setting when the TCAS warning goes off. You are put off for a while, and forget to change your altimeter setting to local QNH with the result that you are now flying 900 ft. below the required altitude. An error has been made. This could have been prevented if the crew had remembered to call out the QNH changeover, after the excitement about the TCAS had been resolved. Once this error has been made, the next option is to detect it at the earliest. This can happen when the PF calls for the completion of the approach checklist. The next checkpoint would be when the radio altimeter is called out and altimeter cross checked. The next stage would be the callout at the Outer Marker, followed lastly by the Landing checklist. In case this error goes undetected then it would be vital to mitigate the consequences of this error by initiating go around at the Decision Altitude, if there are no visual references to land. This is a very mild example of an error and the three step process to prevent incident/ accident. The PNF (PM) by just doing his job of giving the callouts at the correct time would help in neutralizing the error. In an actual case the error could have been introduced by any component of the aviation system involved with the preparation, launch, execution and recovery of the flight. In such cases it may be very difficult, at times, to prevent the error and equally so to detect the error. A 2-crew operation can help in such cases. Our checks, callouts, and procedures have been designed with an aim of detecting and neutralizing errors before they can become problematic. The 2-crew cockpit has the inherent advantage of built in redundancy. It also helps in workload sharing and in ensuring constant crosschecking and monitoring of all actions (by both crew members); monitoring of aircraft trajectory; automation systems and mode status; and aircraft systems and components. To ensure all of this, it is vital to practice CRM. The major rationale for CRM is to enhance crew co-ordination and through this ‘to reduce the frequency and severity of errors’, or error management. Error management can be further enhanced by optimizing human performance in the here and now. What is CRM? NTSB defines CRM as ‘using all available resources – information, equipment and people to achieve safe and efficient flight operations’
Monday, September 1, 2008
On 07 Mar 2007, Garuda Indonesia flight 200 was being operated on a domestic flight from Jakarta to Yogyakarta (Indonesia). The PIC was also the PF. The PIC intended to carry out an ILS approach to R/W 09 at Yogyakarta, and had briefed for the same. ATC cleared the aircraft for a visual approach, with a requirement to proceed to long final and report runway in sight. Although the crew acknowledged the visual approach clearance, they continued with the ILS approach, without informing the controller. Descent and approach were carried out in VMC. At 10.1 miles (initial fix) from the R/W, the aircraft was at 3927 feet (against an altitude of 2500’, as published in the approach chart) at a speed of 283 kts. The PIC intended to make the FAF (6.6 DME) at the correct altitude and thus carried out a steep descent. This did not permit the speed to wash off at the anticipated rate (a tail wind at this altitude also added to the problem). Speed brakes were not selected. He was aware that it was difficult to make the FAF correctly. Flap 1 was selected and thereafter gear was selected down. The PIC called ‘Check speed, flaps 15’. The copilot called ‘Flaps 5’, as the speed was beyond the flap 15 speed of 205 kts. Flaps 5 were selected. The PIC called for ‘flap 15’ twice but the copilot did not select flap 15 as the speeds were beyond the limits. The PIC called ‘Check speed, flaps 15’ again but the copilot did not select flaps as the speed was beyond 240 kts. Also, he did not apprise the PIC of this fact. During the approach the GPWS alerts and warnings sounded 15 times, and the co-pilot also called for a go-around. The PIC continued the approach with flaps 5, and the aircraft attained glide slope close to R/W 09 threshold. The PIC asked the copilot if the landing checklist had been completed, to which he received no response from the copilot. The aircraft crossed threshold at 232 kts (98 kts faster than the flaps 40 landing speed). Ground speed was 235 kts. The aircraft touched down at 221 kts, bounced twice before settling on the runway. Shortly after touchdown, the copilot called for a go around. The PIC selected thrust reverser and continued with the landing. The aircraft overran the R/W, to the right of the centerline at 110 kts. The aircraft crossed a road, and impacted an embankment before stopping in a paddy field 252 metres from R/W 27 threshold. The aircraft was destroyed by the impact forces and post impact fire. 119 of the total 140 occupants survived the accident.
1. Flight crew communication and co-ordination was less than effective after the aircraft passed 2336’ on descent after flap 1 was selected.
2. The PIC flew the aircraft at an excessively high airspeed and steep descent during the approach. The PIC did not go around when stabilized approach criteria were not met.
3. The PIC did not act on the GPWS warnings, and the two call outs by the copilot to go around. (The PIC was 'fixated' on landing the aircraft - was responding to only one stimuli when there were a host of other stimuli seeking his attention. Happens under stress)
4. The copilot did not follow company instructions to take over control of the aircraft when he saw that the PIC repeatedly ignored warnings to go around.
5. Garuda did not provide simulator training to its B737 flight crews covering vital actions and required responses to GPWS alerts and warnings.
Relevant Facts leading to the accident
On 01 Jan 2007, a B737-400 of Adam Air with 102 occupants on board went missing while on a domestic flight from Surubaya to Manado (Indonesia). Last contact with the aircraft was at 14:07, with the flight at FL350. Reports indicate that the flight changed course twice as a result of severe (70 kts) cross wind. The aircraft crashed into the sea killing all occupants. The FDR and CVR were finally recovered from the sea on 27/ 28 Aug, 2007.
Analysis revealed that the autopilot was engaged and was holding 5 deg. left aileron wheel to maintain wings level. Inertial Reference System (IRS) had malfunctioned. Both pilots had become engrossed with trouble shooting the IRS for at least the last 13 minutes of the flight, with minimal regard to other flight requirements. The pilots could not trouble shoot the problem and finally selected the IRS to Attitude. This action disengaged the autopilot, as per design. The auto pilot disengage warning was silenced after 4 seconds but it appears that no pilot was flying the aircraft. After the autopilot disengaged, the control wheel centred and the aircraft began a slow roll to the right. The aural alert, Bank Angle, sounded as the aircraft exceeded 35 deg. bank. The roll rate was momentarily arrested several times, but there was only one significant attempt to arrest the roll. Positive and sustained roll attitude recovery was not achieved. Even after the aircraft had reached a bank angle of 100 deg, with the pitch attitude approaching 60 deg nose down, the pilot did not roll the aircraft to wings level before attempting pitch recovery. The pilots appeared to have become spatially disoriented. Control was thereafter lost; the aircraft had a significant structural failure, and crashed into the sea. The aircraft had recorded a max of 3.5g/ 0.926M/ 490 kts.
Causes as per the Investigation
1. Flight crew co-ordination was less than effective. The PIC, who was also the PF for this segment, did not manage the task sharing; crew resource management practices were not followed.
2. The crew focused their attention on trouble shooting the IRS failure with neither pilot flying the aircraft.
3. After the autopilot disengaged and the aircraft exceeded 30 deg. right bank, the pilots appeared to have become spatially disoriented.
4. The company’s syllabus did not cover complete or partial IRS failure.
5. The pilots had not received training in aircraft upset recovery, including spatial disorientation.
Other Causal Factors
1. 154 recurring defects directly or indirectly related to the IRS between Oct and Dec 2006.
2. Poor Maintenance engineering supervision and oversight.