Summary
Calibrating the Metamax portable metabolic system on the hill, before a test flight.
What did we study and why?
It is said that paragliders crash due to ‘human factors’, poor decision-making or pilot error. However we know very little about the bodies and brains of humans making the decision and errors. So, in Phase I, we first needed to establish some foundations. We started with the body, asking:
- What happens to our heart rates and breathing during flight?
- Why are we tired after flying? How much of the exhaustion we feel is physical?
Who did we study?
We studied four groups of pilots:
Escape Paragliding’s Chris White, Coco Lami, Joshua Sanderson and I flew a total of 9.3 hours in warm, comfortable conditions during the Chabre Open, at a mean altitude of 2236 m.
The awesome Tom de Dorlodot and Horacio Llorens flew 19.3 hours in extreme conditions up to 7458 m during the SEARCH Projects’ 2016 expedition to the Karakorum.
Flyeo’s Malin Lobb, Jim Nougarolles and Guillaume Gensse went through the SIV motions over Lake Annecy.
Finally, with the help of Flymaster, we downloaded all 223 public tracklogs uploaded by pilots using their combined vario and heart rate monitor, the Flymaster Heart-G. We selected out the 81 flights with heart rate data of longer than 20 minutes duration to exclude top-to-bottom flights and to make the data more comparable with the other groups’ flights.
When did we do it?
Summer of 2016 and 2017.
How did we do it?
To measure heart rate and breathing, we used Hexoskins: these were base layers containing fabric stretch-electrodes that gave continuous heart rate tracings, alongside estimates of breathing rate and depth.
The Hexoskins also contained very precise three-axis accelerometers to measure G forces. For Horacio and Tom, we also used oxygen saturation probes to measure what percentage of oxygen was bound to their red blood cells during their extreme altitude flights.
The hardest part was the calculation of energy consumption in flight. We used an instrument called a Metamax, a mask that measured the quantity of oxygen moving in and out of the body with each breath. The difference between the two values was the pilot’s oxygen consumption. Because fuel in the body is ‘burnt’ with oxygen, we could then use the oxygen consumption to estimate the pilot’s energy consumption and physical effort.
We divided the results of each flight into phases for analysis. We picked two five-minute thermal climbs and two five-minute glides from each flight, avoiding the first climb after take-off or the final glide to goal. The take-off phase was defined as the five minutes following the last recorded footfall and the ‘landing phase’ was the five minutes before touchdown.
Results
We showed that pilots had strikingly high heart rates on take off but that otherwise, paragliding was much more about mental rather than physical fitness, though G forces could be high enough to cause loss consciousness and our breathing patterns might have implications for oxygen system design. Of course, our work also had its limitations.
Why oxygen consumption was the right place to start
Oxygen consumption is the closest thing physiology has to a fuel gauge. Every joule of work a body does is ultimately paid for by oxidising fuel, so if you can measure how much oxygen a pilot draws in and how much carbon dioxide comes back out, you can put a number on how hard that pilot is actually working. This matters because free flight is one of the very few sports in which almost everyone assumes the answer without ever having measured it. Ask a group of pilots after a long task and they will tell you, unanimously, that they are exhausted. Ask them what exhausted them and the answers scatter: the heat, the cold, the concentration, the fear, the wrestling with a collapsing wing. Until Phase I there was no dataset that could separate those explanations from one another.
The practical consequence of getting this wrong is not academic. Equipment designers size oxygen systems, harness ventilation and hydration bladders around assumptions about metabolic load. Competition organisers set task lengths and safety margins around assumptions about pilot endurance. Instructors teach fatigue management around assumptions about what tires a pilot out. If the underlying assumption — that flying is physically hard work — turns out to be wrong, then a great deal of well-meaning advice is aimed at the wrong target.
Carrying a laboratory onto a launch site
Field physiology is mostly a logistics problem wearing a lab coat. The Metamax is a research-grade instrument designed to sit on a treadmill trolley in a temperature-controlled room; we asked it to survive being carried up a hill, strapped to a pilot, exposed to sun, wind and altitude, and then flown for the better part of an hour. Calibration had to be done on site, with gas cylinders carried in, and repeated whenever the ambient temperature moved far enough to matter. The mask itself is not comfortable, and pilots deserve enormous credit for flying in it at all — particularly the ones who did so in competition conditions, where every gram and every distraction costs performance.
The instrumented base layers were kinder. Fabric electrodes woven into a close-fitting top give a continuous ECG-quality heart rate trace and a respiratory signal derived from chest and abdominal expansion, plus three-axis accelerometry from a small module at the waist. Because the garment is worn under normal flying clothing, pilots reported forgetting it was there within a few minutes of launch, which is exactly what you want from an instrument that is supposed to record ordinary behaviour rather than provoke unusual behaviour.
Turning tracklogs into physiology
The fourth group in Phase I was different in kind from the other three. Rather than instrumenting pilots ourselves, we worked with a large set of public tracklogs recorded by pilots flying an instrument that logs heart rate alongside position, altitude and vario. That gave us something no field campaign can buy: a large, unsupervised, self-selected sample of ordinary flying, recorded without a researcher standing on the hill.
It also came with obvious caveats, and we treated it accordingly. We excluded flights shorter than twenty minutes to strip out top-to-bottom descents, which are physiologically uninteresting and would have swamped the dataset. We had no way of verifying pilot age, fitness, wing class or the fit of the heart rate strap. What the tracklog set gave us was not precision but breadth — a check on whether the patterns we saw in a handful of closely instrumented pilots also appeared across a much larger population of pilots flying normally, on their own terms, in their own conditions.
How a flight was cut into phases
A paragliding flight is not one activity but several, and averaging across the whole thing hides everything interesting. We therefore segmented each flight into comparable windows: take-off, thermal climbs, glides and the approach to landing. Take-off was defined from the last recorded footfall, which the accelerometer identifies cleanly, and we deliberately avoided the first climb after launch and the final glide to goal, both of which are contaminated by the take-off and landing responses respectively.
Choosing two five-minute climbs and two five-minute glides per flight was a compromise between statistical power and honesty. Longer windows would have blurred the boundaries between phases; more windows per flight would have overweighted the pilots who flew longest. The segmentation is one of the decisions we would most like other groups to challenge, because a different analytic choice might reveal patterns our windows average away.
What Phase I set up for everything that followed
The headline results are set out on the results pages, and they are worth reading in order: heart rate on take-off, physical effort in flight, G tolerance, and breathing pattern. Taken together they point in a consistent direction. The body of a paraglider pilot behaves as though it is under significant stress while doing very little mechanical work. That combination — high arousal, low effort — is characteristic of cognitive and emotional load rather than muscular load, and it is precisely why Phase II turned to cognition, and why our safety teaching puts so much weight on decision-making, group dynamics and preparation rather than on physical fitness.
It is also why we publish our limitations as prominently as our findings. Small samples, field conditions and self-selected participants place real limits on how far these results generalise. They are a first map of a landscape nobody had surveyed, not the last word on it, and we would be delighted to see other groups repeat the work and disagree with us.