A door left open in the rain cost a 747-400 freighter two electrical buses and all its pitot heating: the AAIB report

At 08:13 UTC on 8 February 2024 a Boeing 747-400 freighter lifted off from Runway 09R at London Heathrow bound for Hong Kong. Before the take-off roll was over, already past decision speed, the cockpit alerting system announced the first failure of AC electrical system 4. What followed — the loss of two of the four electrical buses, the loss of all pitot heating, a descent with all three airspeed indications frozen and a 6,240 ft/min dive before landing at Amsterdam — is the subject of the report the UK Air Accidents Investigation Branch (AAIB) published on 3 September 2026 under reference AAIB-29853. The AAIB classifies it as a serious incident: nobody was hurt and nothing was damaged, but the chain of failures began, quite literally, with a door left open in the rain.

The aircraft, registered G-ONEE, is a 747-433 built in 1991 as a passenger aircraft (serial number 24998) and converted into a 747-400SF freighter in 2006 by Israel Aerospace Industries (IAI), powered by four Pratt & Whitney turbofans. It was operating a commercial cargo flight with five crew on board and no passengers. In the right-hand seat, acting as pilot flying, was a 61-year-old instructor commander with 14,600 flying hours, 12,600 of them on type; in the left-hand seat, as pilot monitoring, a captain under line training on one of his first line flights on the 747; and on the jump seat a third pilot acting as safety pilot.

During the take-off roll and initial climb several EICAS messages related to AC electrical system 4 appeared in succession. The autothrottle disconnected and stayed inoperative, and the LNAV and VNAV flight director modes were no longer available. The crew opened the Quick Reference Handbook (QRH) procedure for the loss of AC Bus 4, which warns that only one reset may be attempted per flight. About four minutes later AC Bus 1 also lost power, taking the cockpit voice recorder with it for the rest of the flight. Over the following twenty minutes the pilots managed to manually reset Bus 4 three times, but it kept dropping out until it was permanently unpowered.

With two buses gone, the crew declared a PAN, climbed above the cloud and levelled at FL250 to think. The underlying problem lay in a detail of the 747-400 electrical layout: between them, Buses 1 and 4 power the heating for all the pitot probes, on both sides and the standby. The QRH was unambiguous: avoid icing conditions, because otherwise the captain's, first officer's and standby airspeed indications could all become unreliable. The commander put it plainly to air traffic control: staying out of icing was his main concern.

The crew weighed several diversion options with ATC. Other aircraft on frequency suggested Amsterdam Schiphol, the controller confirmed its weather and, after comparing the runway length available at each option, the pilots settled on Schiphol. They knew they would have to descend through a layer of icing and prepared for it: the safety pilot consulted the QRH "Flight With Unreliable Airspeed" table and set a target of roughly 0° pitch with idle thrust, which should give about Mach 0.84 or 290 kt and a descent rate of around 2,100 ft/min.

What forced them down early was pressurisation. The flight data recorder shows that, with the aircraft level at FL250, cabin altitude began to climb; at that moment the air conditioning packs switched from high flow to normal flow. The crew briefed that they would don oxygen masks if cabin altitude reached 10,000 ft, and the commander began the descent in FLCH (flight level change, which adjusts pitch to hold the selected speed) as the diversion got under way. Cabin altitude started to decrease steadily.

Passing about 12,000 ft the aircraft entered icing conditions. All three airspeed indications began to fall at the same time and the nose dropped. The pilot flying added a little thrust to compensate, but the indicated airspeed kept decaying and the autopilot, chasing in FLCH mode a speed the aircraft was not actually losing, kept pitching down: 7.6° nose-down and 6,240 ft/min. The IAS DISAGREE alert appeared. The pilots recall being about to intervene when, passing 10,000 ft, the aircraft left the icing, the probes cleared and the indicated airspeed jumped by about 85 kt in four seconds. The autopilot pitched up and the pilot flying pulled the thrust to idle, but not in time to avoid the overspeed warning: the peak was 383 kt indicated, 18 kt above the limit, and the warning sounded for 55 seconds. The speedbrakes were not deployed. The commander told ATC they were losing their airspeed indication, asked to continue the descent below 7,000 ft and shortly afterwards declared a MAYDAY. The aircraft broke cloud at 6,000 ft and did not re-enter it.

There was still one more surprise. Schiphol was operating Runway 36R, but ATC offered Runway 06 and the crew accepted. While configuring for the approach a flaps drive caution appeared, which meant another checklist and a higher landing speed; ATC provided radar vectors to give the pilots time to complete it. The landing was uneventful and the aircraft taxied to its parking bay, where the engineers were already waiting.

The crew's training, structured decision-making and use of all crew resources were probably significant factors in avoiding a more serious outcome.

The technical cause of the two lost buses was water ingress into Generator Control Units (GCUs) 1 and 4, located in the E1/E2 rack of the Main Equipment Centre (MEC) beneath the main cargo-deck floor. When they were opened, signs of corrosion and moisture residue were found inside. The question was where the water had come from on an aircraft carrying no cargo on the main deck.

The answer lay in the previous night. While maintenance was carried out, the forward left door, L1, stayed open all night; 4.6 mm of rain fell with an easterly wind of 6 to 10 kt blowing straight into that door. When the crew boarded at about 06:15 they saw water on the door mat and on the floor around it. The subsequent inspection found no blocked drains and no moisture in the inlet fan duct: rain through the open door was the only identifiable source of water.

From there, the investigators found one weakness after another in the barriers that should have contained the water. The L1 door mat, with its built-in drain, was surrounded by a rubber dam, but sealant was missing from its forward-right corner, so water could reach the floor above the MEC. Cargo power drive unit PDU 4 was missing a screw, leaving a hole through which water could drop from the floor onto the drip shield covering the E1/E2 racks. Some of the floor's waterproofing tape was degraded. The drip shield itself had cracks and holes either side of PDU 4, covered by L-brackets whose sealant was also compromised, and the same was true of the shield's end cover plates next to GCUs 1 and 4. Finally, the moisture-barrier curtains that should hang on the front of the E1/E2 racks were not fitted: the operator did not know the conversion manufacturer required them, although the report acknowledges they would not have stopped water entering through the holes in the top of the GCUs anyway. The AAIB could not establish the exact path the water took, but the list of weak points speaks for itself.

None of this was new to the 747-400. The report recalls that there had already been cases of water ingress into GCUs and other contamination of the electronic equipment in the E1/E2 rack, and that these led to service bulletins and airworthiness directives to fit a fibreglass reinforcing overcoat on the drip shield and moisture-barrier curtains on the racks. The most prominent precedent is VH-OJM, a passenger 747-400 which on 7 January 2008, descending into Bangkok, lost Buses 1, 2 and 3 without recovering them and landed with multiple failures in visual conditions; the Australian Transport Safety Bureau investigation attributed the failure to water entering GCUs 1, 2 and 3 from a leak in the forward galley, directly above the MEC. By the time the ATSB was writing its report in 2010, Boeing had received 52 reports of damage to the drip shield or its gutter on other 747-400s, and in November 2009 it had issued service bulletin SB-747-25A3555 to reduce water penetration into the compartment. An engineer who had worked on G-ONEE the night before also recalled that, at another freighter operator, a 747-400 had lost GCUs 3 and 4 to snow melting off the cargo.

The underlying failure the AAIB identifies is one of maintenance: the fibreglass overcoat had been defined as the terminating action, and since then there had been no requirement to inspect it again. The cracks in G-ONEE's shield only came to light because the operator chose to look after the incident. The only scheduled task for that area was a general zonal inspection of the whole cargo deck at each A-check, every 1,000 flight hours, with no specific attention to the area above the MEC; an inspection normally carried out from standing height, from which degraded sealant and tape are easy to miss, and which cannot detect a missing PDU screw, because to see it the unit has to be lifted by hand.

Two safety actions were taken without waiting for the report. On 9 February 2024, the day after the incident, the operator issued a technical notice on ground operations in inclement weather: all main, maintenance-access and cargo doors are to remain closed whenever possible, and closed at all times if no work is being done on the aircraft. IAI, for its part, sent a message to all operators of its 747-400SF on 6 May 2024 reminding them of the service bulletins that improve leakage protection at the MEC and the E1/E2 area.

The AAIB goes further and issues four safety recommendations, two to Boeing Commercial Airplanes for the freighter variants of the 747-400 and two to Israel Aerospace Industries for the 747-400SF. Recommendations 2026-013 and 2026-014 call for a maintenance requirement to repetitively inspect the condition of the drip shield above the E1 and E2 racks of the MEC. Recommendations 2026-015 and 2026-016 ask for the same, as a detailed inspection, for the floor panel seals, the waterproofing tape, the power drive units and the L1 door mat in the area above and around the equipment centre.

On the crew's performance the conclusion is nuanced. There was no QRH procedure for the simultaneous loss of Buses 1 and 4, but the pilots knew they had lost all pitot heating and that airspeed could become unreliable in icing. The rising cabin altitude forced them to descend through the icing; even aware of the IAS DISAGREE alert, they chose to keep the autopilot engaged to reduce workload, with a 0° pitch target in mind, and allowed the automation to lower the nose to 7.6° and 6,240 ft/min. The result was the 18 kt overspeed. The AAIB itself stresses, however, that the crew's training, structured decision-making and use of all crew resources were probably decisive in keeping the incident an incident.

The 747-400 is the best-selling version of the whole family: 694 aircraft delivered between 1989 and 2009, 442 of them passenger variants, as the Boeing 747 entry in this encyclopaedia records. Its great novelty over earlier 747s was a two-crew glass cockpit that cut the dials and switches from 971 to 365, and that is exactly the scenario G-ONEE's crew lived through: a cascade of EICAS messages that two pilots — three, thanks to the safety pilot — had to manage while the aircraft shed systems. It is also an aircraft with a second life as a freighter: the entry recalls that converting retired passenger airframes was for years a predictable business and no longer is. The 747-400SFs and 747-400BCFs still flying, like this 1991 airframe converted in 2006, are therefore an ageing fleet with no replacement, in which the condition of thirty-year-old sealant, tape and shields depends on someone being obliged to look at them. That is the gap the four recommendations try to close; whether Boeing and IAI accept them, and how soon, is what remains open.

Primary document: AAIB Bulletin AAIB-29853, "Boeing 747-433, G-ONEE", Air Accidents Investigation Branch, published 3 September 2026 (© Crown copyright 2026). https://assets.publishing.service.gov.uk/media/6a8c454f08705e34a95daa57/Boeing_747-433_G-ONEE_09-26.pdf

Report announcement: "AAIB Report: Boeing 747-433, G-ONEE", GOV.UK, 3 September 2026. https://www.gov.uk/government/news/aaib-report-boeing-747-433-g-onee

Precedent cited by the report: ATSB Transport Safety Report AO-2008-003 (VH-OJM, 7 January 2008). https://www.atsb.gov.au/sites/default/files/media/3440510/ao2008003.pdf

Model context: the Boeing 747 entry on Above Ten Thousand (history and variants of the 747-400).