Spotting the precursors to accidents requires a multi-prong approach to analysing data

Data can help identify the key causes of a fatal event to target them with counter measures that can save lives. But with accidents becoming rarer, what data do we use?

When discussing risk of fatality there can be a tendency to focus more on fatal accidents with the largest number of casualties. An aircraft that crashes with 15 fatalities may seem a lesser event than a hull loss in which 300 perish. But, both accidents will have causes that need to be understood, and can be acted upon with the right data.

Another challenge with describing risk is that the majority of accidents do not have fatalities because most accidents occur in the ground environment, such as hard landings and runway excursion events. However, the issues that lead to an accident like an overrun, can also manifest themselves in-flight--for example, a training deficiency.

“Sometimes the accident happens on ground, but it can happen in flight, so that is why you need to look at every accident,” says IATA Director of Safety, Rodolfo “Rudy” Quevedo.

Looking at every accident provides good data and that can be measured in different ways. Two common ways that aviation safety can be measured are hull losses per million sectors flown and fatal accidents per million sectors flown. Each is useful, but also may have limitations in terms of helping the industry to fully identify which accidents provide the greatest fatality risk. For example, when using fatal accidents per million sectors as the metric, an accident with one fatality receives as much weight as an accident with 300 fatalities.

Another method that is able to deliver great insights is referred to as fatality risk.

Fatal fascination

Fatality risk is also being used by the Commercial Aviation Safety Team, which is a partnership initiative between the United States government’s Federal Aviation Administration (FAA) and industry stakeholders. ICAO is another organization using fatality risk.

Fatality risk measures the exposure of a passenger or crew to a catastrophic accident where all people on board the flight are killed. Quevedo explains that the fatality risk approach to measuring accidents takes the view that an accident with all-onboard perishing is one accident (1.0), while an accident where a quarter of the passengers and crew die is a quarter of an accident (0.25). The calculation of fatality risk does not take into account aircraft size or how many were on board. What is measured is the percentage of those on-board who were killed. This is expressed as fatality risk per millions of flights.

There were a total of four accidents, three of which involved fatalities. The two accidents that each involved 50% fatalities equate to one full loss equivalent. That means that in total there were two full-loss equivalents out of three million sectors flown, which equates to 0.67 full-loss equivalents per million sectors. In other words, the exposure of all passengers and crew who flew on those sectors to a catastrophic accident was 1 in 1.5 million flights.

Quevedo emphasizes that the other methods of analysing accident data are not being dropped, but that fatality risk can help the industry focus in on those causes that can be countered to have the biggest impact in terms of saving lives. The accidents with the greatest fatality risk are those that occur in flight, such as Controlled Flight Into Terrain (CFIT) and Loss Of Control-Inflight (LOC-I). They are the two principle causes of deaths.

Quevedo says: “We are using [fatality risk], along with other metrics, more succinctly to be more data driven in our approach. For example, many organisations may start a project on training if they think there is a deficiency in pilot training for a specific thing.” IATA is working with airlines, their systems and data is to see what they say are the deficiencies behind accidents

IATA is including the fatality risk approach in its Safety Report, which is published annually in April. The report says that fatality risk provides a good baseline for comparison between accident categories. This can be seen with LOC-I which is known to have a high fatality risk, but a low frequency of occurrence.

With databases of events, deficiencies and counter-measures, investigators can measure the effectiveness of mitigation actions that are put in place. “If we increase pilot training for an area, which could prevent loss of control, then we should see loss of control accidents diminishing,” says Quevedo.

The pilot training that is the mitigation action for LOC-I could be bank angle exceedance. Measuring every time an aircraft rolls past 45 degrees should show that their frequency is going down.

Unknown others

However, counter measures can only be identified if the data is broad enough. Understanding any safety issue from as many angles as possible is key to developing successful preventative measures.

There are parts of the world where collecting the facts is more difficult. This has resulted in the third largest area with fatalities being the Other category, for unknowns. In accidents, investigators will know that there were fatalities, but sometimes not enough is known about the event to arrive at a probable cause. “We don’t know enough about the accident to say it is loss of control, or even runway excursion, we just know some people died. It is the area where we are working to see what can be done about this,” explains Quevedo.

The goal, therefore, in developing countries is to build accident investigation capabilities to get that much needed data. According to the ICAO 2014 safety report, the area of least improvement was accident investigation. ICAO’s Deputy Director Safety, Air Navigation Bureau, Catalin Radu, points to nations lacking adequate legislation and regulations related to investigations. “The fact that an accident investigation authority should be fully independent from the administrative and financial perspectives was also a challenge for many States to quickly address,” says Radu.

However, accidents are becoming so rare there is an argument for not having a national investigatory body that spends most of its time having nothing to investigate. Instead, regional bodies could be used to collect that all important data. “One of the discussions is, do we have regional organizations that have the expertise and can do this work?” says Quevedo.

The idea of regional investigation organizations in parts of the world where resources are limited is also being promoted by ICAO. Radu said that ICAO has been encouraging the establishment of what it calls Regional Accident and Incident Investigation Organizations. These are to help, “address the fact that related costs can at times run to many tens of millions of [US] dollars,” explains Radu. Where regional organizations have not come about, Radu explains that, so far, the, ”quality of a major investigation has not been directly impacted to any major extent due to,” international cooperation filling in any capability gaps.

More data means more safety. While some parts of the world will need to address their data deficiencies, other regions will be able to use the new statistics to address their training or procedural problems that can be the cause of fatalities. Last year saw no jet hull losses. With the introduction of another risk analysis tool, the hope is that all this will make us even safer.