The Gimli Glider
On the 23rd of July 1983, Air Canada Flight 143 ran out of fuel at FL410 – 26,000 feet halfway through a 2,829 kilometre (1,768 mile) flight from Montreal to Edmonton. The Boeing 767-200 carrying 61 passengers managed to glide to safety to the Gimli Industrial Park Airport in Manitoba, a Canadian prairie province.
What went wrong? The sequence of events leading to the incident is somewhat convoluted.
The first link in the chain of events took place almost three weeks prior to the fault, on the ground in Edmonton. Boeing had issued a service bulletin for the Fuel Quantity Indicator System (FQIS) to be checked on all 767s. From an operational standpoint, the plane could not be flown if the FQIS was not functional at all, but with partial functions, the plane was considered operational but the fuel quantity needed to be measured with a float stick. The Boeing 767 C-GAUN was given a routine check by an Air Canada technician, who found that testing the three fuel quantity indicators caused them to go blank. However, during a later check the indicators worked fine.
The same technician saw the issue again, although he wasn’t aware that it was the same plane. This time he traced the problem to a faulty circuit breaker. Disabling it meant the back-up circuit breaker kicked in and the gauges worked. He pulled the breaker, tagged it as “inoperative” and apparently left a note in the log book which was not very clear.
The next day, C-GAUN flew to Montreal. Here, a maintenance worker saw the note in the logbook and checked the circuit breaker. While waiting for the fuel truck, he decided to try resetting the breaker and the fuel quantity indicators went blank. The fuel truck arrived and he got to work, forgetting about the faulty circuit breaker that he had reactivated.
This action was considered to have “contributed significantly” to the accident.
The return flight to Edmonton was a different crew. The captain noticed the blank fuel gauges and stated to the crew that they would need to perform a drip test to check fuel levels. He decided to load the full amount of fuel needed to fly directly to Edmonton with an en route stop at Ottawa to verify fuel levels.
This fuel requirement is expressed as a mass, in this case 22,300kg of fuel.
There was an estimated 7,682 litres of fuel remaining in the tanks.
The maintenance crew worked out how many litres of fuel were needed to make up 22,300kg of fuel, then subtracted the 7,682 litres on board and then used the fuel gauge on the refuelling truck to fill the aircraft tanks with the remaining required litres of fuel.
Canada was at this time changing from imperial to metric. The Boeing 767 was the first plane in the fleet to measure fuel in kilograms rather than pounds.
The maintenance crew had a multiplier of 1.77 for converting from litres. Somehow, no one noticed that this figure was for a conversion to pounds, not kilograms.
The maintenance crew calculated the amount of fuel needed using a factor of 1.77 pounds/litre rather than 0.8 kg/litre and reported that the plane had 22,300kg on board. It actually held 22,300 pounds, which is just over 10,000kg.
In the past, when fuel was calculated manually, a flight engineer’s duties included checking the fuel load. Flight engineers were a thing of the past on this 767, as a Presidential task force, under Ronald Reagan, had determined that aircraft could be built to be operated by two pilots instead of three, if the tasks previously given to the second officer (flight engineer) were either fully automated or handled by ground staff. Responsibility for ensuring adequate fuelling had passed to the maintenance branch. But because these men were not trained to calculate fuel, they assumed the pilots would make sure it was done properly.
The problem was neither of the pilots was trained in this technical task. Safety procedures had failed to keep pace with new technology. As the investigation later concluded: “Air Canada … neglected to assign clearly and specifically the responsibility for calculating the fuel load in an abnormal situation.”
The flight crew checked the figure but they only checked the arithmetic, not the conversion factor, so they came up with the same result. They manually entered the fuel amount into the flight management computer as 22,300kg. The computer tracks fuel consumption by subtracting the fuel burned from the total amount.
At Ottawa, a further drip-stick test was made and there was a further chance to spot the error. The Captain was told that the aircraft had 11,430 litres of fuel on board. He converted this using the figure that the ground crew had given him in Montreal: 1.77/litre and came to the conclusion that they had 20,400 kilos of fuel remaining. Still no one realised that the previous crew had used pounds rather than kilos. The aircraft had less than half the fuel remaining that the captain thought it did, 9,144 kilos of fuel, and nowhere near enough to complete the flight to Edmonton.
The flight crew verified the remaining fuel against the computer and took off.
When the first low fuel pressure warning sounded, the Captain believed that the left fuel pump had malfunction. He turned it off, asked for a a divert to Winnipeg and began his descent. Another low fuel pressure warning sounded, this time for the right side. Then the left engine failed, rapidly followed by the right engine.
As Pearson began gliding the big bird, Quintal “got busy” in the manuals looking for procedures for dealing with the loss of both engines. There were none. Neither he nor Pearson nor any other 767 pilot had ever been trained on this contingency. Pearson reports he was thinking “I wonder how it’s all going to turn out.” Controllers in Winnipeg began suggesting alternate landing spots, but none of the airports suggested, including Gimli, had the emergency equipment Flight 143 would need for a crash landing. The 767’s radar transponder had gone dark leaving controllers in Winnipeg using a cardboard ruler on the radar screen to try and determine the 767’s location and rate of descent.
Pearson glided the 767 at 220 knots, his best guess as to the optimum airspeed. There was nothing in the manual about minimum sink – Boeing never expected anyone to try and glide one of their jumbo jets. The windmilling engine fans created enormous drag, giving the 767 a sink rate of somewhere between 2000 and 2500 fpm. Copilot Quintal began making glide-slope calculations to see if they’d make Winnipeg. The 767 had lost 5000 feet of altitude over the prior ten nautical (11 statute) miles, giving a glide ratio of approximately 11:1. ATC controllers and Quintal both calculated that Winnipeg was going to be too far a glide; the 767 was sinking too fast. “We’re not going to make Winnipeg” he told Pearson. Pearson trusted Quintal absolutely at this critical moment, and immediately turned north.
Only Gimli, the site of an abandoned Royal Canadian Air Force Base remained as a possible landing spot. It was 12 miles away. It wasn’t in Air Canada’s equivalent of Jeppensen manuals,but Quintal was familiar with it because he’d been stationed there in the service. Unknown to him and the controllers in Winnipeg, Runway 32L (left) of Gimli’s twin 6800 foot runways had become inactive and was now used for auto racing. A steel guard rail had been installed down most of the southeastern portion of 32L, dividing it into a two lane dragstrip. This was the runway Pearson would ultimately try and land on, courting tragedy of epic proportions.
The co-pilot suspected that the Captain had not seen the guardrails nor the crowds of people. Knowing they had only one chance at landing, he decided to keep his mouth shut. Pearson came in high and fast but managed to use the rudders to lose speed and altitude in the same way as you would in a glider or small aircraft. It worked.
The rally spectators were startled to see a huge aircraft bearing down on them, silent except for the rushing of wind against its body. People scattered as quickly as they could, but only the friction between the aircraft nose and the ground as the partly extended nosewheel collapsed, brought the aeroplane to rest in front of them. The time was 2038 hours. Just 17 minutes had elapsed since Pearson had started flying a powerless 767 from 28,500 feet to a safe landing.
In 2008 the Boeing 767 was decommissioned. Its final flight is documented on YouTube:
Amazing plane, amazing story. And for people like me, flying an American plane using gallons in Europe where fuel is measured in litres, it is not a bad reminder always to check the conversions.