Mozambique Presidential Tu-134 C9-CAA

A presidential return flight

On 19 October 1986, Mozambique's presidential Tupolev 134A-3, C9-CAA, carried President Samora Machel and other senior officials on a journey to Zambia. It left Maputo for Lusaka in the morning, took on fuel there and continued to Mbala. The aircraft reached Mbala at 09:00 UTC and remained there until the evening. It received no fuel at Mbala before the return flight.

C9-CAA left Mbala at about 16:38. Nine crew members and 35 passengers were aboard. The five flight deck crew members were Soviet citizens; the four cabin crew members were Mozambican. The crew planned to reach Maputo at 19:25. Their navigation log named Beira as the alternate airport, but no flight plan was filed for the sector. The computed departure fuel mass was 10,384 kg.

The descent clearance

At 18:46:19, the radio operator contacted Maputo Information and reported the aircraft at Kurla, maintaining flight level 350. He gave the arrival estimate as 19:25, an endurance of four hours and 48 people aboard. Investigators established that the actual occupant count was 44 and the endurance was three hours 45 minutes. Maputo advised that there was no reported traffic at that level and no expected delay for an ILS approach to runway 23.

The crew confirmed that they would continue to the Maputo VOR. At about 19:02 they reported the top of descent, and the controller asked them to report at 3,000 feet or with the runway lights in sight. The aircraft began descending. Before descent, the autopilot pitch channel had been disengaged, while its roll and yaw channels remained engaged. The flight recorder initially showed only small lateral deviations from the required track.

A turn eight minutes early

At 19:10:41, while passing about 19,000 feet, C9-CAA began a right turn from 184 degrees magnetic towards 221 degrees. It banked between ten and 15 degrees and completed the turn at 19:11:39. The navigator reported 100 km remaining during the turn. When the captain questioned the change of direction, the navigator attributed it to the VOR indication.

The South African board concluded that the navigator initiated the turn through the Doppler heading selector and autopilot. It placed the manoeuvre about eight minutes before the point at which the aircraft should have turned to intercept the 045 degree radial for Maputo runway 23. The resulting heading resembled the heading the captain expected for the approach, and the board considered that this helped explain his acceptance of the turn.

The board considered the Matsapa VOR in Swaziland the probable source of the misleading indication. Matsapa transmitted on 112.3 MHz and Maputo on 112.7 MHz. Its preferred explanation was that a pilot inadvertently selected Matsapa on the No. 1 receiver and later selected the Maputo ILS frequency. It also considered a temporary selection of Matsapa for a cross-check. These were reconstructions of the likely frequency selection sequence.

Checks and position information

The cockpit arrangement made spoken cross-checking particularly important. The navigator sat behind the co-pilot and could not readily see the VOR frequency selectors on the front panels. He could select which receiver supplied his indicator, but he could not select its frequency himself. The board found that the prescribed spoken frequency checks and identification of navigation aid coding were absent from the recording.

During descent, the co-pilot listened to a Russian language broadcast of music and news on the HF radio. The captain discussed earlier flights with low fuel and the allocation of drinks to crew members. The board described these as distractions from monitoring the flight path. It considered the navigator's approximate distance reports almost certainly to have come from the Doppler equipment. The DME readings would have given accurate distance information.

By 19:17:21 the captain was expressing concern that Maputo was missing from view. The crew attributed their failure to receive expected navigation indications and see the airport to an electrical failure on the ground. The board found that the aircraft's position prevented reception of the expected ILS signals. It also found that the Maputo runway lights were working.

Below 3,000 feet

At 19:18:24, C9-CAA reached 3,000 feet and reported maintaining that altitude. The recorder showed a continuing descent. At 19:18:46 the controller cleared an ILS approach to runway 23, gave transition level 40 and a pressure setting of 1017, and asked the aircraft to report established on the 045 degree radial. The radio operator replied that the ILS was out of service.

The controller answered affirmatively and cleared a visual approach to runway 05. The crew continued looking for runway lights while descending. The navigator gave distances of 25 to 30 km and later 18 to 20 km. The captain referred to lights on the right, while the co-pilot said the runway was unlit. Requests to check the runway lights brought further visual approach clearances.

The report found misunderstandings in the exchanges between the aircraft and the tower. Normal radio transmission and reception continued, but the two sides had difficulty conveying and interpreting the intended messages. The crew remained uncertain of their position and continued below the assigned altitude. The board concluded that they lacked the visual contact with the ground required for the descent they were making.

The terrain warning and impact

At 19:21:02 the ground proximity warning sounded. It continued for 32 seconds. The captain reacted verbally and the aircraft made a slight nose-up pitch change, but the descent continued. The board described the average descent rate in this final period as about 480 feet per minute. The crew were still discussing cloud, runway lights and the unavailable navigation indications.

The flight manual required a decisive climb with take-off power following a terrain warning over hilly or mountainous ground. It required the same response when the terrain beneath the aircraft was unknown. The board concluded that compliance, even seconds before impact, would have prevented the aircraft from flying into the ground. It considered a climb to a safe altitude sufficient to permit reassessment of the navigation information.

C9-CAA struck the ground at 19:21:39 UTC, or 21:21:39 local time. The site was at Mbuzini, approximately 35 nautical miles west of Maputo and just inside South Africa. Its coordinates were 25 degrees 54 minutes 41 seconds south and 31 degrees 57 minutes 26 seconds east. The impact elevation was 666.75 metres, or 2,187 feet.

Deaths and survival

The aircraft was destroyed. The report classified eight crew members and 26 passengers as fatally injured. The flight engineer and eight passengers received serious injuries, and one passenger received minor injuries. Machel was among those killed. Ten of the 44 occupants survived the impact. Nine of those ten had been seated in the rear right section of the cabin, while the flight engineer had been in the cockpit section.

Survivors recalled being told at about 19:00 to fasten their seat belts for landing. They described the aircraft as operating normally until impact.

One seriously injured passenger died of his injuries on 5 January 1987. The board retained him in the serious injury category for statistical uniformity, so its accident table continued to show 34 fatalities.

The board's findings and the dissent

The South African board attributed the crash to the flight crew's failure to follow instrument approach procedures, their descent below safe and assigned altitudes in darkness and some cloud without visual ground contact, and their failure to respond to the GPWS alarm. It found the aircraft and relevant equipment serviceable before impact. It also found insufficient fuel to reach Beira with the required reserves. Its calculation gave about 2,400 kg remaining against a diversion requirement, including reserves, of 4,730 kg.

Mozambique and the Soviet delegation dissented. Mozambique disputed the depth of the inquiry into a decoy VOR and the absence of a flight test replicating the proposed reception of Matsapa. Its analysis called the explanation for the turn unresolved and asked the three states to continue investigating. The Soviet delegation rejected the board's analysis and conclusions, argued that Matsapa could not explain the flight path and signal reception, and attributed the diversion to a deliberately operated false beacon.

The Soviet analysis also said that the crew treated the terrain warning as false because they believed they were near Maputo and doubted the radio altimeter. The board replied to the delegations in Part II, amended its draft in response to some Mozambique comments, and maintained its findings on the cause and the beacon theory. The final report was signed on 2 July 1987.

Recording and recommendations

The investigators recovered the MARS-BM cockpit recording and translated the Russian conversation with participation from the three countries. They combined it with the air traffic control recording. A failed digital time signal and an unavailable transmission timing parameter prevented direct synchronisation, so they used the terrain warning as the common event and adjusted the flight recorder time by nine seconds. The synchronised data gave an elapsed flight time of two hours 44 minutes 22 seconds and a corrected take-off time of 16:37:17.

The board recommended correct use of standard radio terminology and recurrent training, particularly where crews or controllers worked in a foreign language. It also called for review of the monitoring of crew compliance with operating procedures, maintenance of navigation aid integrity, and an increase in continuous cockpit voice recording from 30 minutes to at least one hour. It considered that a longer recording might have helped explain the inadequate fuel reserves.