Articles
Crash landing: G-YMMM came down well short of the runway at Heathrow
At lunchtime on 17 January 2008 a British Airways Boeing 777 aircraft, powered by two Rolls-Royce RB211 Trent turbofan engines, was on its final approach to Heathrow airport from Beijing. Everything about the flight of passenger aircraft G-YMMM had been routine until 57 seconds before touchdown when at a height of 720ft the starboard engine experienced ‘rollback’ – a sudden uncommanded reduction in thrust.
Seven seconds later the same happened with the port engine. With 27 seconds to go before touchdown and both engines now at ‘idle thrust’ power, the airspeed started to drop below the 135 knots required for a safe approach. Quick thinking by the flight commander averted what might otherwise have been a disaster. He altered the aircraft’s flap setting to reduce drag and the plane managed to land 110m inside the airport’s perimeter fence, although still 330m short of the runway.
Nevertheless the effects of that landing were violent and destructive. The nose landing gear and both sets of main landing gear collapsed, with that on the starboard side sheared right off. But there were no fatalities and only one serious injury – a broken leg suffered by a passenger.
How much worse it could have been had been demonstrated as far back as 1958 when another Boeing aircraft, a B52 bomber, suffered a sudden reduction in power in three engines as it approached the US air force base at Ellsworth in South Dakota. The aircraft came down 3,500ft short of the runway, resulting in fatal injuries to three of the crew.
The connection between the two incidents was not just their superficial similarity but their fundamental cause – ice collecting at a crucial ‘choke point’ so that it inhibited the flow of fuel to the engines. In the case of G-YMMM the sequence of events was detailed in a report published two years after the incident by the Department for Transport’s Air Accidents Investigation Branch (AAIB).
The report stated that when the plane took off it carried 79,000kg of fuel distributed between a centre tank and two main wing tanks – one for each engine. During its flight the aircraft experienced an unusually low external total air temperature of –45°C which in turn produced a minimum recorded fuel temperature of –34°C, although neither of these was outside its ‘operating envelope’.
The flight was atypical because of a combination of two factors – a relatively low cruise fuel flow, followed by a high fuel flow with the fuel still at a low temperature as the plane manoeuvred to land. The AAIB report said this made the flight unique among 175,000 other flights with which it was compared in a data-mining exercise.
These circumstances caused a build-up of ice on the inside of the pipes of each engine’s fuel system. The ice was then carried along by the enhanced fuel flow until it reached each engine’s fuel oil heat exchanger. This unit is designed to force the fuel to flow through narrow parallel channels instead of a single pipe so that it can be warmed through a heat-exchange process with hot oil from the engine.
Crucially the entrances to that array of tubes projected slightly ahead of the main body of the unit so that they were not directly heated. So the ice that impacted them did not melt but built up to block them and starved the engines of fuel at a critical moment with all the near-disastrous consequences that followed.
In the case of G-YMMM what happened was understood and an engineering solution was found to ensure that those precise circumstances cannot recur – a simple redesign that cut back the entrances to the tubes so that they became flush with the main body of the unit. But achieving a more fundamental understanding of ice formation in aviation fuel is still a subject for intensive research.
A summary of current knowledge was provided by a symposium that took place recently at the HQ of the Institution of Mechanical Engineers. It was evident that, although the formation of such ice is routine, indeed inevitable, is recognised as such and is generally well managed, it remains a phenomenon that cannot be predicted or modelled with complete accuracy.
The AAIB’s report into the G-YMMM incident, which featured prominently at the event in a presentation from the organisation, reinforces that impression.
Water, the report states, will occur naturally in jet aviation fuel in one of three states. These are: dissolved – where the water and fuel bond at a molecular level; entrained (also known as suspended) – where the water exists as tiny droplets in the fuel;
free – where the water is neither dissolved nor entrained and generally settles under gravity on the bottom of the fuel tanks.
The precise proportions in which these different states will exist, though, are not fixed. Instead, as fuel becomes colder, dissolved water will become either entrained or free. As a general rule, the AAIB’s report states, the dissolved water content of aviation fuel in parts per million is approximately equal to the temperature of the fuel in degrees Fahrenheit.
A rough calculation shows that the actual amount of water in a plane full of fuel is therefore very small. The fuel loaded onto G-YMMM before take-off probably contained three litres of dissolved and two litres of entrained or free water, although small further amounts may have been left over from previous fuel loads or have entered the system in flight as atmospheric water vapour.
As the dissolved water reverts to an undissolved state – a natural consequence of the decrease in temperature experienced by a plane in flight – it becomes susceptible to freezing. However, as the AAIB’s presentation explained, the behaviour of ice in aviation fuel is not consistent over the entire temperature range it experiences, but instead exhibits a series of different characteristics.
Initial ice formation occurs as soon as the temperature dips below zero in the range –1 to –3°C, but then in the range –8 to –20°C it becomes ‘sticky’ and adheres to the interior of the metal fuel pipes. Maximum adhesion is at –12°C. Below –20°C, stickiness diminishes as the ice becomes more crystalline.
Besides this variable behaviour of the ice, another complicating factor is evident. Although the ice forms in the fuel system that is part of the aircraft’s structure, its effects, if any, will be manifested in the engines, which are designed and made by a different company.
This situation is acknowledged by Mark Reid, fuel and oil systems specialist with aeroengine manufacturer Rolls-Royce. But, as he points out, that does not absolve the organisation of any requirement to address the issue. “We still have a responsibility to make sure that what we do in our system doesn’t contribute to the situation,” he says.
Until the G-YMMM incident, that involved ensuring compliance with requirements laid down by the European Aviation Safety Agency and the US Federal Aviation Agency, which stipulated safe and continuous operation of jet engines with a proportion of water as ice within the fuel up to a concentration of 260ppm.
But G-YMMM changed all that and made the industry realise that it also had to develop standards to deal with the sudden ‘snow-shower’ release of a much larger concentration of ice into the fuel flow. As Reid candidly admits, that sort of event was a “completely new type of scenario that had never been considered before”.
Then a second though much less dramatic incident in which a Boeing 777 suffered a loss of power to an engine through ice in the fuel occurred later the same year. On 26 November 2008, the plane with the registration N862DA operated by Delta Airlines experienced a reduction in thrust in its right engine on a flight from Shanghai to Atlanta 55 minutes after it had completed a step climb that had involved a major increase in fuel flow from 6,900 pounds per hour to 10,900pph. Fuel flow was restricted to a maximum of 5,000pph for 23 minutes.
The obvious similarity with the G-YMMM incident – a reduction in thrust after an increase in fuel flow – has to be offset against at least two major differences. The first is the much greater time lag involved between the increase in fuel flow and the thrust reduction, and the second the fact that only one of the two engines was affected.
The requirement, therefore, says Reid, is for standards that will define the ‘snow-shower’ phenomenon and also prescribe universal methods of testing engines for it. So far, he confirms, no such standards exist although the regulatory bodies have issued requirements relevant to specific new engine development projects – known by the European Aviation Safety Agency as certification review items.
But the foundations for the creation of such standards are now being laid. The agency has initiated three major relevant pieces of research since the G-YMMM incident. Two of them have already been completed – a project called Water in Aviation Fuel in Cold Conditions and another known as A Survey of Fuel Anti-Ice Additives for Civil Aviation. The third, Ice Accretion and Release in Fuel, got under way at the end of last year and is still in progress.
Someone closely involved in the first of those initiatives was Dr Joseph Lam, a specialist in computational fluid dynamics with Airbus in the UK. Lam says that ice formation and growth in fuel are similar to frost formation and growth in the atmosphere, which may open up the possibility of useful knowledge transfer from the world of atmospheric physics. “We may tap into the research on frost formation and growth in the atmosphere to help us further our understanding of ice formation and growth in fuel,” he says.
At a more immediately practical level, though, Lam says that Airbus Fuel & Inerting Research & Technology has been developing a fuel dehydration technology for next-generation aircraft to eliminate as far as possible the fuel system water drain task. That procedure is time-consuming and costly. The organisation demonstrated a proof of concept at industry conferences in 2010 and 2011.
In addition he says that Airbus is working closely with a partner in Russia in the development of fuel dehydration techniques.
Relevant research is also under way in universities. A major centre of activity is Cranfield University where several researchers are exploring the field in pursuit of PhDs. One of them is Solange Baena, a contributor to the IMechE symposium, who says she is seeking to establish mathematical tools to analyse the behaviour of ice in aviation fuel systems when it meets a ‘mesh type’ barrier. Funding for her work has come from Airbus and the Engineering and Physical Sciences Research Council.
Baena dislikes the use of the term ‘random’ to describe such behaviour – she says that would imply the situation cannot be modelled. Instead she says that there are “a lot of variables”.
Her work is not focused on the interface between the fuel system and the engine that caused the G-YMMM incident but on a mesh that sits much further back with the primary aim of protecting from random debris the pump that drives the fuel through the system. She is due to submit her thesis to the university later this year.
It will add to the growing body of knowledge through which the industry is seeking to get to grips with a phenomenon that, although it occurs on just about every flight, has the potential to spring a surprise on one of the most sophisticated and safety-conscious areas of engineering expertise.