Flight Scoring Part Three

In the previous article, I said transfer functions would be the place to start in flight scoring, but to understand why, we need to talk briefly about triangles. We all remember trigonometry back in school. Whether you were taught on a chalkboard, whiteboard, or on a projector screen, the core of it was the same, three sides, and three angles all adding up to 180°. At the heart of all of this are the sin(x), cos(x), and tan(x) functions. For flight scoring we don’t need to delve too deeply into each, we just need to know that they are all transfer functions, and that they need to be seen as continuous curves, not as a digital yes/no signal. For trigonometry to work perfectly, I need to be able to get a different sine value for different inputs; my triangle wouldn’t be very triangular if the sine value for 60° and 80° inputs were the same.

With flight scoring, you need to know that two landings, with all else being equal, one 1500ft along the runway and the other 2000ft along, can be scored differently. Here we need to take advantage of the curves of an analogue-esque signal. Digital signals here are similar to the event level thresholds used in ‘traditional’ Flight Data monitoring (FDM), where any value between 1.000 and 1.999 might be a ‘level 2’, and anything 2.000 or above might be a ‘level 3’. This is where the ‘resolution’ comes in.

Here I have calculated Y=sin(x), but in different resolutions; the blue line takes the measurement every 5 degrees, and the yellow takes the measurement every 45 degrees. The important thing to note here is how the 30 degree line, with its three levels, seems eerily similar to the way level 1, 2, and 3 events are monitored in FDM. This is what I call the “1.000=1.999 problem”

Applying Trigonometry

Now, you’re not here to have me teach you about trigonometry, you’re here to find out how to improve your FDM. Thankfully, there is a solution in sight, flight scoring. Flight scoring should be seen as an enhancement for your ‘traditional’ FDM programme, rather than a replacement. Event monitoring is still vital, and event-based statistics still have their place, but their use in detecting trends, particularly for an individual crew member, can be far exceeded by flight scoring.

However, is a taxi speed exceedance equally as bad as a landing outside of the marked runway area? No,, but false event equivalency can be found in nigh-on all FDM programmes. False event equivalency mostly creeps in when FDM programmes rely heavily on what could be called the ‘traditional statistics’ (top 10 level 3s, top event rates from the last 3 months etc). Tracking top events is still very important, and should definitely still be done, Flight Scoring is an enhancement of the programme, not a replacement after all, but these statistics should not be the be-all-and-end-all. Safety departments should be particularly careful with these types of statistics, because they can draw focus away from potentially more important issues that are making their way up, but aren’t triggering events yet.

When we are limited to triggering certain event levels, and if, like in many operations, the safety department doesn’t have all the resources it needs to function the best it can, the ‘1.000=1.999 problem’ can crop up in a more insidious way. Let’s say that we have two flights with bad landings, these flights might trigger the same events at the same levels, and with pilots who have never triggered these serious events before. Now, tracking events, reviewing event statistics over the whole operation, or even tracking individual crew member’s statistics wouldn’t have flagged up anything of interest, but flight scoring could.

A good FDM system will be taking hundreds of different measurements per flight, from the maximum taxi speed, to how far from the runway centreline the aircraft touched down. There are many grey areas to consider when it comes to flight safety, but one we can all agree on is that the aircraft should touch down on the runway centreline. The measurement in L3Harris’ Flight Data Connect system for this is called “Distance From Runway Centreline At Touchdown”. Now, the further the aircraft touches down from the centreline, the higher this landing would score under flight scoring. For example:

An important thing to note here, is that not all Key Point Value (KPV) measurements need to be compared to the same Y-scale. The taxi speed exceedance might have a maximum score of 0.5 for example, whereas factors like this graph, or hard landings, might not have a maximum value; a score can go as high as is needed for each individual measure. All an operator would have to do here is set their transfer function, defining the curve shape, the scores, and a few points on the curve, and all flights could then have their respective measurements, compared to this curve, and their scores calculated.

Approaching your flight data in this way eliminates the ‘digitalness’ of the event-based statistics, adding dimensionality. Flight Scoring can bring other benefits too, improved trend detection, the ability to track individual crew member’s strengths and weaknesses, and vastly improved inter-operator comparisons. To find out how, join us for part 4.

Flight Scoring Part Three

Applying Trigonometry

Read Part One

Read Part Two