Air Canada Flight 143

A fuel indication fault in Edmonton

Air Canada flight 143 was scheduled from Montreal to Edmonton with an intermediate stop at Ottawa on 23 July 1983. Captain Robert Pearson and First Officer Maurice Quintal were flying Boeing 767 C-GAUN, fleet number 604. Six cabin crew and 61 passengers were aboard. The aircraft's fuel gauges measured weight in kilograms, while its fuel measuring sticks were calibrated in centimetres. Air Canada's other aircraft used fuel gauges reading in pounds. Fuellers supplied fuel by volume in litres.

The sequence began during a service check in Edmonton the previous night. All three cockpit fuel quantity indicators were blank. The digital fuel gauge processor had two channels, either of which normally could supply the fuel indications. Technician Conrad Yaremko found that pulling the channel 2 circuit breaker restored the gauges through channel 1. He put yellow tape marked inoperative around the breaker and a second piece marked see log book above the fuel indicators. He recorded the defect in the journey log and deferred it through the airline's computerized system.

With one processor channel inoperative, the Minimum Equipment List required confirmation of fuel loading by a measuring stick or tender uplift and availability of flight management computer fuel quantity information. Yaremko arranged a drip check. Captain John Weir then flew the aircraft from Edmonton to Montreal through Ottawa on the morning of 23 July. The three cockpit gauges operated normally during that flight. The distinction between a failed processor channel and blank cockpit gauges became important when the next crew took over.

The gauges went blank again in Montreal

Before the new crew arrived, avionics technician Ouellet examined the journey log and the tagged circuit breaker. His training did not seem to agree with the log entry, so he attempted a built-in test of the processor. He pushed the channel 2 circuit breaker back in before the test. The cockpit fuel gauges went blank. He decided that the processor needed replacement, but no replacement was available in Montreal. One was expected to be available that night in Edmonton.

On returning to the cockpit, Ouellet was distracted by the fueller's arrival and forgot to pull the channel 2 breaker again. The tape collar remained around the breaker even though it was now pushed in. When Pearson boarded, he assumed that the collared breaker had been deactivated and that it was the breaker for the cockpit fuel gauges. He had also received the impression from a brief conversation with Weir that the aircraft had already been operating with blank gauges. Those impressions reinforced each other.

Pearson consulted the Minimum Equipment List. Item 28-41-1 required at least two of the three fuel gauges to work for dispatch, and item 28-41-2 required one processor channel. Pearson testified that an attending technician had told him that Maintenance Central had authorized operation with blank gauges. The inquiry found that no such authorization had been given and questioned whether the conversation occurred. Pearson decided that he could safely fly after the quantity of fuel had been confirmed with measuring sticks. The aircraft departed Montreal, and later Ottawa, in contravention of the Minimum Equipment List.

Fuel volume became an incorrect weight

A drip check measured the depth of fuel in each tank. The crew and maintenance personnel then used tables to convert the centimetre readings to litres. To obtain fuel weight in kilograms, they also needed the correct weight conversion factor. The factor supplied in Montreal was 1.77 pounds per litre. Everyone involved treated the result of multiplying litres by that factor as kilograms. A kilogram conversion factor for the fuel was about 0.8, with variation according to temperature.

The first drip readings corresponded to 3,758 litres and 3,924 litres, a total of 7,682 litres already aboard. Multiplying by 1.77 produced 13,597 pounds, which the participants treated as 13,597 kilograms. They subtracted that figure from a minimum requirement of 22,300 kilograms, then divided the supposed shortfall by 1.77. This produced a request for about 4,916 litres. The fueller suggested rounding the uplift to 5,000 litres.

The inquiry calculated that the existing fuel actually weighed about 6,180 kilograms. Against the minimum requirement, approximately 20,036 litres should have been added, about four times the amount loaded. The flight plan also included 300 kilograms for taxiing, bringing its actual fuel requirement to 22,600 kilograms. Using that requirement would have increased the uplift further. Both flight crew and maintenance personnel participated in the incorrect calculations. The inquiry found that they lacked training and experience in converting the drip readings into fuel weight.

Pearson had decided to load enough fuel in Montreal for the journey through to Edmonton and to check it again in Ottawa. The Ottawa check repeated the conversion error. Maintenance personnel supplied a total volume of 11,430 litres. Pearson asked for the conversion factor and received 1.78. His calculation produced more than 20,000 units, consistent with the quantity he expected after the flight from Montreal. The check therefore accepted the same underlying error, and flight 143 continued with far less fuel than its crew believed.

Both engines stopped during the diversion

Shortly after 8:00 p.m. Central Daylight Time, the cockpit instruments warned of low fuel pressure at the left fuel pump. Pearson immediately began a descent from 41,000 feet and diverted towards Winnipeg, then about 120 miles away. Within seconds, warnings appeared for the right main tank. Within minutes the left engine failed, followed by the right. At that point the aircraft was at 35,000 feet, about 65 miles from Winnipeg and 45 miles from Gimli.

Loss of the main electrical supply left the pilots with a standby magnetic compass, artificial horizon, airspeed indicator and altimeter. These worked, but the compass was positioned high above the instrument panel between the pilots and was difficult to read accurately because of parallax. The normal flight management information was gone, and there was no standby vertical speed indicator. Quintal used successive altitude readings and distances supplied by Winnipeg controllers to estimate the descent profile while also working through emergency procedures.

The crew concluded that Winnipeg could not be reached. At 8:32 p.m., in consultation with air traffic control, Pearson redirected the aircraft towards Gimli, then about 12 miles away on the western shore of Lake Winnipeg. The cabin crew had instructed the passengers in emergency procedures. Gimli was a former military airfield with a 7,200-foot runway. Its far end had been adapted for drag racing, and drivers and their families were staying beyond the racing strip in tents and caravans.

One approach and a rapid evacuation

Without engine power, the crew could not use flaps or slats to control the descent. Pearson and Quintal discussed a sideslip to lose height and speed as they approached Gimli. Pearson, who had gliding experience, performed the manoeuvre on final approach. The aircraft touched down within 800 feet of the runway threshold.

Quintal had delayed lowering the landing gear to reduce drag during the glide. When selecting the normal gear lever produced no extension, he searched the Quick Reference Handbook for a free-fall procedure. He could not find it in the landing gear section or its index. He then selected the alternate extension switch, allowing the gear to fall under its own weight. The main gear locked down, but the nose gear did not. He searched the hydraulics material for further guidance and ran out of time. The procedure was at the end of that section.

The nose gear collapsed as it contacted the runway, bringing the lower nose into contact with the concrete. That contact damaged the aircraft but added friction and helped it stop before reaching the people beyond the drag racing strip. The fuel tanks were dry. All 69 occupants survived, and the inquiry reported no significant physical injuries.

Evacuation was quick with 61 passengers aboard, but the nose-down attitude made the rear slides steep and left their ends above the ground. Cabin crew also expected the overwing ramps to be directly in front of the exit windows. When they could not see them there, they concluded that the ramps had failed to inflate and redirected some passengers to the front and rear exits. Their manual had not explained the ramps' position, and their training had given them the wrong impression.

Hardware, organization and fuel checks

The Board of Inquiry traced the processor malfunction to an inductor coil in the channel 2 power supply. The failure reduced the logic supply and prevented the expected transfer to the good channel. The inquiry identified a design failure in that switchover protection. The processor fault had been intermittent, and its built-in test system became unreliable during the failures. The precise timing of the internal failure remained unknown.

The inquiry also found deficiencies in Air Canada's organization. Responsibility for manual fuel calculations in abnormal situations had not been clearly assigned. Manuals were unclear, training was insufficient, and the history of fuel indication problems was poorly communicated to personnel handling the aircraft. The inquiry identified failures of communication at both operational and management levels, including coordination between Flight Operations and Engineering during introduction of the 767.

By the final report of April 1985, Air Canada had changed the drip charts to convert centimetre readings directly to kilograms. It had assigned abnormal fuel calculations jointly to maintenance personnel and flight crew, requiring independent readings and calculations followed by comparison. Fuel slips used the term Fuel Weight Conversion Factor, the airline had introduced an automatic uplift check through ACARS with a manual alternative, and it had increased its stock of spare processors. Honeywell had improved the processor's design and quality.

Lockwood recommended additional safeguards, including second-person verification of required drip checks, clearer and standardized manuals, and instruction that the aircraft's Minimum Equipment List and its qualifying conditions were binding. He also recommended improvements to communication between Flight Operations and Engineering, access to maintenance histories, and corporate organization. These recommendations addressed the successive failures that allowed the aircraft to depart with blank gauges and insufficient fuel.