Business Analytics Blog

Business Analytics Blog

Opinions expressed on this blog reflect the writer’s views and not the position of the Capgemini Group

How safe is it to fly by Maija Antila and Nigel Lewis

On 8th March 2014 Malaysian Airlines flight MH370 to Beijing disappeared within less than an hour after taking off from Kuala Lumpur in the “dead space” between Malaysian and Vietnamese Air Traffic Control (ATC). Just before the last radio message, ACARS, the on-board system that sends performance information, had stopped functioning and the transponder which communicates a plane’s position had been switched off. Without these mechanisms to communicate with the outside world the aircraft appeared to have disappeared.

It was discovered that there was one data source that could give insight into the path that MH370 had taken that night. There were automatic ‘pings’ that were sent between the Inmarsat satellite and the plane and the Inmarsat satellite and the ground station. The ‘ping’ is sent once an hour to record whether a plane is still logged on but it does not contain any location information. This data confirmed that the plane had continued flying between six and seven hours after the last communication.

A large amount of data analysis and new way of modelling was involved in trying to figure out the path that the plane took after the communication ceased. The analysis used the time between the sending and receiving of the ‘ping’ signal to calculate two arcs of possible locations of the aircraft, a northern and a southern corridor. It used the velocity of the aircraft relative to the satellite and the change in signal frequency known as the Doppler Effect (the reason why police or ambulance sirens sound different when coming towards you compared to going away from you) and compared the data with six other Boeing 777 flights flying in different directions on the same day. The analysis had good match with the southern corridor. Knowing the direction the plane was flying and the time before it went down it was concluded that the aircraft likely went down in the South Indian Ocean, somewhere west of Australia.

This terrifying, mysterious and extremely sad incident has evoked the question about how safe it is to fly. I am particularly interested in this as in a couple of weeks’ time I am due to fly to the Indian Ocean. Should I cancel the holiday and stay home instead?

Over the past 20 years the probability of being killed in a flight accident when flying with one of the 78 major world airlines has been 1 in 4.7 million. Fly with one of the top-39 airlines and the probability goes down to 1 in 19.8 million. Flying with the safest airlines further reduces the probability of accidents dramatically. For example, British Airways has had no fatal accidents over the past 37 years and they have over 40 million passengers annually.

Most of the accidents are caused by a combination of events, as reported for example in National Research Council’s book Improving the Continued Airworthiness of Civil Aircraft, which alone would not cause the accident but together lead to a crash. According to research on accidents, often it is a combination of bad weather, minor technical issues, the flight being behind schedule so it is hurrying, pilot being tired and unable to think sharply and the pilots not being comfortable with each other.

A typical accident involves on average seven consecutive human errors. The errors that cause planes to crash are invariably errors of teamwork and communication. In his book Outliers Malcolm Gladwell highlights the role of the cultural background of the pilots and the significance of having a low ‘power distance’ between the pilot and the co-pilot. Historically, crashes have been far less likely to happen when the co-pilot, not the pilot, is flying. This is because then the second pilot (in these cases the more senior pilot) is not afraid to speak up. In accident investigations, psychologists play an important role analysing the recordings from the cockpit.

Understanding of the causes of accidents has led to more efficient training and consequently the safety of flying has increased significantly. The past five years of passenger flight accident data show that the probability of being killed in a flight accident has been 1 in almost 7 million. Choose one of the safest airlines and you are more likely to be killed by a vending machine than in a plane crash.
A closer look at the accident data over the past five years would suggest that, statistically, flying on a Friday might minimise the probability of an accident (data not normalised).

Data source:

Also, this data suggests that an intercontinental flight may be safer than flying within a continent. Actually, a third of all passenger flight accidents happened on flights within Asia.

Data source:

This analysis has given me reassurance that the risk of something very bad happening on my flight to the Indian Ocean (Saturday, Europe to Africa, British Airways) is very small and I am ready to take it. The risk of getting killed in London traffic is probably higher.

Follow Maija Antila @maija_antila

About the author

Nigel Lewis
Nigel Lewis
Nigel leads the Capgemini Consulting’s 35 strong Business Analytics team, which delivers analytical, operational and strategic modelling solutions to clients. He has 18 years consultancy experience as well as 8 years experience in the UK gas industry. Nigel has successfully managed complex projects in both the public and private sector, including capacity modelling, simulating supply chain operations, strategic business modelling to support future policy decisions, and implementing complex demand forecasting systems. Nigel is currently focussing on the development of Capgemini’s customer analytics and analytics advisory services.
3 Comments Leave a comment
Great article - I am now getting a bit worried, as I am going to Asia this summer, and will be flying within the continent! I also remember a talk at a conference a number of years ago about flight safety - they mentioned a number which was the acceptable number of crashes per passenger mile. This number came from the 50s, in an era when there were fewer flights. I seem to remember that back then it came to about 1 crash every 20 years. But now with the increased number of flights, the number works out at a crash every 2-3 years!!!
Maija should be listed in the right hand column listing the authors.
It is something we are working on Lucy.

Leave a comment

Your email address will not be published. Required fields are marked *.