ATM Modernization, Business & GA, Commercial

Safety: Descent to Disaster

By David Evans | March 1, 2006
Send Feedback

The approach mode was never armed, and as a result, neither the localizer nor the glide slope was captured. The captain, who was the pilot flying, spent the approach attempting to capture both, with what appears to be complete focus on only one of them at any particular time. Eventually, the ground proximity warnings sounded and were ignored until too late–the airplane crashed during the attempted go-around, which was initiated just 100 feet above the ground.

The captain, with more than 16,000 hours flying experience, was killed. The first officer and four of the 21 passengers were injured, and the airplane, a Bombardier CRJ-100, was destroyed.

Back to the Beginning

Details of the June 22, 2003, Brit Air crash while attempting to land on runway 26 left at Brest, France, are contained in the 140-page report issued in December 2005 by the BEA, the French accident investigation authority.

As is sometimes said of history, roughly, "We know the end, but it is not so simple to go back to the beginning." In this case, however, BEA did go back to the beginning, the design of the instrument array in the cockpit, and offered some useful thoughts.

First, the details of the accident: it was at night, and thunderstorms were in the area, prompting the crew to deviate around them, with acknowledgment from air traffic control.

The issue of complacency and/or crew fatigue is evidenced in this BEA report observation: "The incomplete arrival briefing, prior to carrying out the pre-descent checklist, may indicate the onset of routine (crew returning to base, fourth flight of the day)."

Then the set-up, the failure to arm the approach mode, as set forth in the BEA report:

"Switching through `HDG’ [heading] mode, changing the navigation source to `VOR,’ and activating the ILS frequency all correspond to actions prior to arming of the `APPR’ [approach] mode. However, at the end of said sequence, this mode had not been armed. In fact, had it been armed at that moment, the result would have been immediate capture of the localizer beam. In addition, there was no verbal announcement relating to … the pushbutton or FMA [flight mode annunciator] display.

"Two explanations may be offered for non-arming of the mode: the captain could have forgotten to do so at that moment, given the number of actions he had to perform in a short time and, perhaps, the attention he was paying to the exchanges with the controller, or perhaps he was waiting for the airplane to steady at 2,000 feet, and with his attention focused on the managing the descent, he subsequently forgot.

"However, though it may seem improbable, it cannot be excluded that a rapid action on the pushbutton was not taken into account by the system.

"For his part, the co-pilot did not check whether the `APPR’ mode was armed," the BEA report stated. "He also could have been troubled by repetition of the [controller’s] messages."

The sequence of events from that point forward might be described as a descent to disaster. The airplane passed under and then above the glide slope, and then drifted under it until crashing. There was a continued drift to the left. Here the BEA was particularly trenchant:

"The aircraft’s drift was not detected by the crew for about two minutes. The method of display employed for LOC [localizer] and GLIDE [glide slope] information on the PFD [primary flight display] is not limited to the CRJ-100 and -700 alone, and meets the requirement of the certification regulations.

"However, separating the two items, even within a single screen, widens the visual scan that the pilots must perform when making an approach and tends to make them focus on one of them. It is possible that the crew would have detected the aircraft’s drift earlier if both sets of information had been presented in an interconnected manner."

There was a belated effort to correct the leftward drift, and the autopilot was disconnected at this point, too. The co-pilot said, "Come right" twice, the second time just around the moment when the terrain alert announced "one hundred" [feet].

At about 93 feet, the co-pilot said, "I’ve got nothing in front," and the captain announced, "Go around." Engine thrust increased, and pitch attitude changed from -5 to +0.6 degrees, although about four seconds later. About 15 seconds before impact, the "Too Low Terrain" alert sounded, but by that time the airplane’s speed had decayed to 115 knots and impact, within sight of the airport, was unavoidable.

As BEA noted in its report, "The numerous `glide slope’ and `sink rate’ … alarms … indicated that the approach was not stabilized."

Among its many recommendations, BEA said that combining the localizer and glide slope information on one instrument, rather than on two instruments displayed on the same screen, should be considered. On this particular aircraft, the PFD has an artificial horizon at the top of the screen and an HSI (horizontal situation indicator) below. The glide slope is shown as part of the artificial horizon, and the localizer is shown as part of the HSI. The HSI display also may be too "busy" for clarity, and it appears to be partly hidden by the yoke.

Scan Issues

This may not be a particularly good design. Unlike the single picture on the head-up display, which was on the aircraft but not used for this approach, the head-down display requires the pilot to integrate the separately displayed information from two instruments in his head.

As one pilot remarked, "If you have to search for what you need to instinctively assimilate, then the data display is alienating. None of this will be apparent as a deficiency when just monitoring a coupled approach."

But in this case, because the APPR mode wasn’t armed, a manual precision approach had to be flown, with all critical tasks and timings up to the crew. While the regulations permit separate display of glide slope and localizer information, one has to wonder why it is permitted in this day and age, when integrated displays are clearly within the state of the art and instrument panel real estate is at a premium anyway.

The complete BEA report of the incident is available at www.bea-fr.org/docspa/2003/f-js030622a/pdf/f-js030622a.pdf.

Receive the latest avionics news right to your inbox