Aviation Investigation Report A06A0096

On 12 July 2006, an Air Canada Embraer 190-100 (registration C-FHIU, serial number 19000037) was being operated as Flight ACA1156 on a scheduled flight from Edmonton, Alberta, to Toronto, Ontario, with five crew members and 81 passengers on board. At 1011 mountain daylight time, the aircraft commenced its take-off. During rotation, the crew noticed that the aircraft pitch response was different than normal. The aircraft successfully climbed away and the flight continued and made an uneventful landing in Toronto.

History of Flight

The flight crew was scheduled for a 10-hour duty day, operating two segments. The first segment, from Toronto/Lester B. Pearson International Airport (CYYZ) to Edmonton International Airport (CYEG), was flown by the captain. A 45-minute turn-around was scheduled, followed by a return to Toronto.

On arrival in Edmonton, company maintenance staff had to power down^{Footnote 1} the aircraft to clear a secondary power distribution assembly (SPDA) fault message. Because aircraft servicing was to be completed with only battery power available, the captain had to oversee the refuelling operation and supervise the servicing of the lavatory system. The captain completed these tasks, while the first officer was completing the pre-flight walk-around inspection. After the power-down exercise, the SPDA fault message was cleared and the aircraft was declared serviceable.

For the return flight to Toronto, the captain was the designated pilot not flying (PNF) while the first officer was the pilot flying (PF). Company standard operating procedures (SOPs) require that the PNF calculate the take-off performance data using an on-board laptop computer commonly referred to as an electronic flight bag (EFB).^{Footnote 2} The thrust setting and take off speeds are then to be verified by the PF, copied onto the operational flight plan, and then entered into the flight management system (FMS).

After completion of the walk-around, the first officer calculated the preliminary take-off performance data using the captain's laptop. The captain's laptop was used because the first officer's laptop power cord was defective. The captain's laptop had been left plugged in to ensure continued battery charging. However, the first officer had to reach over to the captain's side of the cockpit because the power cord on the captain's side was not long enough to reach to the first officer's side. All preliminary load information was entered. However, an error in the fuel weight was introduced when the first officer entered the weight of the fuel on board (FOB) at the time, which was 3700 kg, instead of entering the planned fuel for departure, which was 10 200 kg

The first officer then transcribed the resulting take-off performance data to a blank area at the upper right corner of the operational flight plan (see Appendix A). These data included the take-off weight (TOW), thrust setting, decision speed (V1), rotation speed (Vr), take-off safety speed (V2), flaps up speed (Vfs), and the stabilizer trim position. The transcribed data were 41 700 kg, 84.9, 137, 137, 140, and 186 with a stabilizer trim of 1.3 UP, respectively. Had the planned FOB weight of 10 200 kg been used, the resulting figures would have been 47 600 kg, 90, 149, 149, 151, 200 and 1.3 UP trim (see Table 1). When the captain returned to the flight deck, he assumed his PNF duties and proceeded to enter the preliminary performance figures into the FMS.

Table 1. Performance data

* indicates fuel in thousands of kilograms

Approximately 15 minutes before departure, the captain was advised that the flight service director (FSD) was not on board. He then contacted the Station Operation Control (STOC) in an attempt to locate the FSD. Shortly thereafter, the FSD arrived.

The crew members received their air traffic control (ATC) clearance, and entered the navigational data into the FMS. As part of the pre-flight fuel check, they compared the fuel gauge indication (10.2 tons) with the operational flight plan required fuel (10.15 tons). They verified that the fuel sheet indicated the proper flight plan revision number, and completed the ACARS^{Footnote 3} pre-flight fuel check. The SOPs did not require that they compare the actual FOB with the FOB entry on the laptop computer take-off page. Consequently, no discrepancy was noted.

Shortly after initiating the before-start checklist, a flight attendant (FA) advised the flight crew that water was overflowing from a coffee maker. While the first officer communicated with the FA over the interphone to resolve the water overflow problem, the captain was advised by ATC that the departure runway had been changed, and that they would now be departing from Runway 12.

Because there had been a problem with the first officer's laptop power cord, the captain completed the changes using his laptop. He entered the new runway, temperature and altimeter setting, and recalculated the take-off performance data. By design, the system automatically transferred the fuel figure that had been initially entered (3700 kg), and used this figure for the new calculation. The new performance data generated were then compared to the previously calculated data and, because of the similarities, were accepted as valid. The captain did not identify either the incorrect fuel weight, or the incorrect take-off weight presented on the laptop take-off page. He then entered the new thrust and take-off values into the FMS.

When the crew members called ATC for clearance to push back from the gate, they were unable to contact either ground or tower control because those two frequencies were unserviceable. The crew members were eventually able to contact departure control. They received push-back clearance 10 minutes after the scheduled departure time. After engine start and release of the ground support crew, the crew requested, and received, taxi clearance for a departure from Runway 12.

During taxi, the crew received, and the first officer reviewed, the final load data on the ACARS. The final load data values were compared to the operational flight plan values listed at the bottom left of the operational flight plan, and accepted, because they were within the prescribed SOP tolerances. At no time were the final load data values compared to the EFB values that had been transcribed to the top right corner of the operational flight plan, nor was this required by SOPs.

The first page of the operational flight plan lists the planned block fuel, estimated zero fuel weight and estimated take-off weight at the bottom left corner. A designated area situated immediately right of these figures is used to transcribe the final load data. There is no specific area on that same page to transcribe the performance data calculated on the EFB. Crews normally transcribe these data to the top right corner of the page.

At 1011 mountain daylight time,^{Footnote 4} the aircraft took off from Runway 12. The aircraft was rotated at a speed of 140 knots, at a rate of 1.5 to 2 degrees per second, in a smooth and continuous motion. The actual lift-off occurred at 159 knots, eight seconds after the first up-elevator input to the flight control. During rotation, the crew noticed that the aircraft pitch response was different than normal. The aircraft felt as if it were out of trim and slow to respond.

The aircraft was not equipped with any device that would have provided the crew with an accurate and timely indication of inadequate take-off performance. Presently, there are no such devices certified for installation or use on civil aircraft. In recent years, considerable effort has gone into the development of a reliable means to detect inadequate take-off performance. Transport Canada (TC) created a team to continue to explore ways of building a take-off performance monitoring system using emerging technologies as proof of concept.

Once above 10 000 feet above sea level (asl), the crew members reviewed the performance data on the EFB and noted the discrepancy. They immediately advised company dispatch. The rest of the flight was uneventful. On arrival in CYYZ, the crew filed an Air Safety Report (ASR), as is required by the company to address safety concerns.

Weather

The Edmonton International Airport weather recorded at 1000 was as follows: winds 040° true at 10 knots, visibility 15 statute miles (sm), few cloud layers starting at 1800 feet above ground level (agl) with a ceiling at 11 000 feet, temperature 15°C, dewpoint 10°C, and altimeter setting 29.75 inches of mercury.

Flight Crew

The crew for this flight was current and qualified and was experienced in airline operations. The captain had completed his training on the Embraer in April 2005. The first officer had completed training on the Embraer in October 2005.

Electronic Flight Bag

The Air Canada Embraer fleet is equipped with Class 1 EFB for determining performance calculations. This EFB is a stand-alone computer that does not share connectivity with aircraft systems. The EFB was introduced into line operations when the Embraer aircraft entered service at Air Canada. The EFB is a laptop computer using a Microsoft Windows®-based software application called Legato. The application was developed in house by a team of aircraft performance staff and technical pilots of the Air Canada Flight Operations Embraer team, and was approved for operational use by TC.

Take-off performance is calculated in Legato on the take-off page. This page was designed with a two-step calculation process to increase awareness of preliminary versus final load figures. The first calculation assumes that the preliminary figures for the upcoming flight are entered using the estimated zero fuel weight (EZFW) and the planned fuel on board (FOB) for the flight. The final calculation is completed when the crew receives the final load data.

Standard Operating Procedures

The SOPs for the Embraer were developed by a core team of experienced company check and training pilots, mainly from the Air Canada Regional Jet program. They used procedures provided by the aircraft manufacturer, SOPs from other Air Canada fleets, and procedures from other airlines operating the aircraft. As the SOPs were being developed, they were trialed in a simulator for appropriateness.

TC monitored the development and ultimately approved the procedures. This was done in accordance with the Canadian Aviation Regulations, specifically Section 725.138 of the Commercial Air Service Standards (CASS) and the associated guidance of Section 745.138. For the EFB, Commercial and Business Aviation Advisory Circular (CBAAC) 231, dated 20 July 2004, was used as guidance. CBAAC 231 is based on Federal Aviation Administration (FAA) Advisory Circular (AC) 120-71A. The Canadian standards and guidance material used for the SOP approval provides direction and advice that is primarily for content.

Other Occurrences

On 14 October 2004, a Boeing 747 cargo aircraft crashed on take-off from Halifax, Nova Scotia,^{Footnote 5} when the crew attempted a take-off with less than the required thrust and lower-than-required take-off rotation speed. All seven crew members suffered fatal injuries. The crew had used the incorrect take-off performance data that had been calculated on the on-board laptop computer using the aircraft weight from the previous take-off.

In June 2006, the TSB issued Recommendation A06-07 to TC stating that the Department of Transport, in conjunction with the International Civil Aviation Organization, the Federal Aviation Administration, the European Aviation Safety Agency, and other regulatory organizations, establish a requirement for transport category aircraft to be equipped with a take-off performance monitoring system that would provide flight crews with an accurate and timely indication of inadequate take-off performance. Although the installation of such a system in aircraft would not prevent data insertion errors from happening, it was seen as a physical defence that would assist the crew in identifying take-off performance data errors. Following this recommendation, TC formed a working group with the objective of exploring ways of building a prototype Take-off Performance Monitoring (TOPM) system.

On 10 December 2006, a Boeing 747-400 with 563 passengers and 15 crew members on board took off from Paris-Orly Airport, France, also using incorrect take-off performance data. Substantial damage occurred to the lower aft fuselage area when the aircraft over-rotated during the take-off. The runway was sufficiently long that the aircraft was able to become airborne and return for a safe landing. In this instance, the crew had used the aircraft zero fuel weight (ZFW) instead of the take-off weight (TOW) to calculate the take-off performance data.

Following the Paris-Orly occurrence, France's Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile (BEA) organized a working group to study the issue of data insertion errors leading to take-off performance occurrences. The working group will be analyzing data from similar past occurrences. Taking part in this working group will be the Direction Générale de l'Aviation Civile (DGAC) of France, airlines and the Laboratory of Applied Anthropology (René Descartes University - Paris V). The aim is to gain a better understanding of the cause(s) of this phenomenon in order to issue appropriate safety recommendations to reduce or eliminate this risk.

There have been several reported accidents and incidents that demonstrate that crews throughout the airline industry continue to attempt take-offs using incorrect take-off performance references (see Appendix B). In the occurrences identified in this report, the procedural defences that were in place did not function effectively and, on several occasions, this has resulted in substantial aircraft damage and loss of life. The frequency of occurrences shows that the risk of crews using incorrect take-off performance references remains high because the cockpit procedural and technical safety defences to prevent these errors have been either inadequate or absent.

The aircraft was serviceable for the flight and there were no environmental factors that were considered contributory. The analysis will therefore focus on the safety defences that were not effective in preventing the crew from attempting to take off using incorrect take-off performance references.

During the initial performance calculation, an error in the weight of fuel on board (FOB) for take-off was introduced. Therefore, the performance data generated was not correct. This resulted in the use of a lower-than-required thrust setting for take-off, V speeds that were inappropriate for the weight of the aircraft, and less-than-optimum performance and handling characteristics.

The turn-around time scheduled for the flight in CYEG was 45 minutes. During the course of a normal day, flight crews expect to have to deal with unforeseen or changing circumstances. The requirement to power down the aircraft on arrival and the EFB power source anomalies were particularly significant. In this case, the situation created an increased workload, particularly for the captain, and disrupted the normal procedural flow. While readying the aircraft for departure, both crew members deviated from SOPs. Whenever there is a deviation from normal procedures, there is more vulnerability to errors. Other non-standard events that would have added to the crew's overall workload management were the defective laptop cord, the absent flight service director, the overflowing coffee maker, the late change to departure runway, communication difficulties with ATC, and the later-than-scheduled gate departure.

SOPs are the primary safety defence to ensure that there is a correct, standard, and safe outcome for each phase of flight. To achieve this, the procedures must be sufficiently robust to withstand the daily challenges that crews face during normal operations. SOPs must be effective in circumstances where the crew is working under pressure, interrupted or distracted, facing non-standard situations, and in periods of high workload. Consequently, there must be human performance considerations during the design, development, and use of these critical defences. In this occurrence, the procedures in place failed to ensure that correct generation and use of take-off performance data occurred.

SOPs required the PNF to calculate the take-off performance data and the PF to verify the data before transcribing the data to the operational flight plan. These data could then be entered into the FMS. However, on this day, this was not done because both pilots deviated from SOPs and the PF calculated the take-off data.

Had the verification of the preliminary performance calculations been accomplished correctly, the discrepancy likely would have been identified at this point. The FOB field would have indicated 3700 kg instead of 10 200 kg and the estimated take-off weight (ETOW) field would have indicated 41 700 kg instead of the operational flight plan ETOW of 47 780 kg.

Although the SOPs required that the crew transcribe the EFB performance generated data to the operational flight plan, there was no specific location on the operational flight plan to transcribe the take-off performance figures. Therefore, the crew wrote them down on the top right corner of the page. Since the planned figures were located in the opposite corner, at the bottom left of that same page, it was difficult to compare the calculated performance figures and the planned figures. Had there been a specific area, adjacent to the planned figures, to transcribe the calculated performance data, the discrepancy between the planned take-off weight and the take-off weight used for generating the performance figures would have been easier to identify. It is more likely that the six-ton difference between the planned and calculated take-off weight would have caught the attention of the crew.

The fuel check was then completed. No discrepancies were noted since, by that time, the aircraft had been refuelled and the fuel gauge indication of 10.2 tons matched the operational flight plan required fuel of 10.15 tons. There were no specific requirements to verify the actual FOB against the FOB entry on the EFB take-off page during the fuel check procedure or the pre-flight procedure.

In this occurrence, the crew took off using incorrect take-off performance references. This was a result of a combination of the crew deviating from SOPs and the operational flight plan form design that prevented easy identification of discrepancies. The procedural defences that were in place did not function effectively and this could have resulted in substantial aircraft damage and loss of life.

Findings as to causes and contributing factors

1. The normal flow of activities to prepare for the outgoing flight was interrupted, and the flight crew deviated from the standard operating procedures (SOPs) while readying the aircraft for departure.
2. An incorrect aircraft weight was used to calculate take-off performance data. This error was not detected, and resulted in the crew conducting the take-off with lower-than-required thrust and speed references.
3. There is no area on the operational flight plan for transcribing calculated performance data. This prevented easy identification of the take-off weight discrepancy.

Other Finding

1. A take-off performance monitoring system could have provided the crew with an accurate and timely indication of inadequate take-off performance.

Safety action taken

On 03 August 2007, the TSB issued Aviation Safety Advisory A06A0096-D1-A1 (Use of Incorrect Take-Off References) to Transport Canada (TC). The Aviation Safety Advisory informed TC of the Paris-Orly 10 December 2006 accident, and indicated that TC may wish to coordinate its efforts with those of the Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile (BEA) to accelerate safety action to mitigate the risk of crews using incorrect take-off data references.

On 21 December 2007, TC responded stating that it is continuing to work towards realizing the intent of TSB Aviation Safety Recommendation A06-07. A team of participants from diverse branches of TC and from the National Research Council (NRC) have met and agreed to a project work plan.

Through the International Air Transport Association (IATA) Safety Group, Air Canada has engaged the manufacturers to develop automated flight management system (FMS) take-off data entry gross error checking capabilities.

This report concludes the Transportation Safety Board's investigation into this occurrence. Consequently, the Board authorized the release of this report on 31 January 2008.

Appendix A - Operational Flight Plan

Image

Appendix B - Take-off Performance Incident/Accident Occurrences

Appendix C - Glossary

AC

Advisory Circular

ACARS

Aircraft Communication Addressing and Reporting System

agl

above ground level

asl

above sea level

ASR

Air Safety Report

ATC

air traffic control

BEA

Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile

CASS

Commercial Air Service Standards

CBAAC

Commercial and Business Aviation Advisory Circular

CYEG

Edmonton International Airport, Ontario

CYYZ

Toronto/Lester B. Pearson International Airport, Ontario

DGAC

Direction Générale de l'Aviation Civile

EFB

electronic flight bag

ETOW

estimated take-off weight

EZFW

estimated zero fuel weight

FAA

Federal Aviation Administration

FMS

flight management system

FOB

fuel on board

FSD

flight service director

IATA

International Air Transport Association

kg

kilograms

NRC

National Research Council

N1

thrust

PED

portable electronic device

PF

pilot flying

PNF

pilot not flying

sm

statute miles

SOPs

standard operating procedures

SPDA

secondary power distribution assembly

STOC

Station Operation Control

TC

Transport Canada

TOPM

Take-off Performance Monitoring (system)

TOW

take-off weight

TSB

Transportation Safety Board of Canada

Vfs

flaps up speed

Vr

rotation speed

V1

decision speed

V2

take-off safety speed

ZFW

zero fuel weight

°C

degrees Celsius