Hacker News Reader

Comment by JumpCrisscross | Hacker News Reader

> On Feb. 22, 2013, Morris attached an accelerometer to his helmet, then launched his body down a 1,500-meter track at the sliding center in Whistler, British Columbia, which is considered the fastest track in the world and was a venue for the 2010 Winter Olympics.

Bobsleigh has been a feature of the Winter Olympics since 1924 [1]. Yet it might have taken until 2013 for someone to measure its g force on its athletes' heads.

One would think that when something is developed to the point of an international sport, it's certainly been measured. Been tested. But the disturbing or revealing or inspiring truth is, most of our normality has not been scientifically considered.

[1] https://en.wikipedia.org/wiki/Bobsleigh_at_the_Winter_Olympi...

View on HN

> it might have taken until 2013 for someone to measure its g force

To be fair, measuring this sort of thing was pretty hard in the past, when cheap miniaturized sensors and memory didn't exist.

I expect we will see more and more activities being re-evaluated now that tech is easily wearable.

You could do it with a weight on a spring, and a little pen attached to the weight.

That won’t handle the number of degrees of freedom a bobsled uses.

Just thinking this through, wouldn't three weights on springs get you all the data? Or some sort of membrane containing a liquid than ruptures when a particular amount of force is felt in any direction?

The problem with three axes is that the registration of the largest deviation by the pen happens for the axes independently.

Yet its still valid data in that its trivial to calculate the theoretical maximum peak accel based on the possibly independent peak accels of the individual axes.

If in theory the max peak acceleration could never exceed 2 G then there's no point in continuing the study. If the theoretical max peak accel could be 57 G then there's a point in continuing the study.

The whole problem is irrelevant anyway. The problem isn't one individual sled journey, the problem is thousands of competitors taking thousands of rides over many years, so "tens of millions". They are competitive with each other having small std deviations in performance so each sled ride is roughly the same. Tiny reliable cameras have been a thing for a century as have mechanical timers. Simply bolt a half pound telemetry package to every sled that snaps a single random point of data via camera or some weird linkage to clutches and springs. Then run a data analysis and make statistical predictions. You wouldn't even need a 1960s mainframe to run the statistical analysis. This would have been something near BAU for mid-century aeronautical engineers, for example. I'm sure something similar was implemented during the test program for the F-86 or maybe P-51.

Peak is not enough - you need to know time impulse. A 10G accel over 1 sec has much different effect than a 10G over a nanosecond.

And the measuring devices must also handle this nuance. A liquid has inertia, so won't register quick impulses, and the things it measures might not be relevant to what bio systems are interested in.

True. How about suspending a little metal bullet in a gel, chosen such that the bullet will break through the gel if the G-force is more than a certain threshold. This will give a good indication if the forces are within an acceptable range.

Reminds me of shockwatch for packages.

This kid of stuff is incredibly hard to use for accurate results. Gels (and all materials) are not perfect in any sense, and you'd need significant effort modeling any material you used - viscosity under all sorts of motions (since nothing is nicely linear), the gel itself is sloshing, compressing, has different lag times...

It's possible in theory, but in practice would not be useful.

Look at what planes (Air Force has lots of R&D) did for the past century to see the state of the art at each point and the methods used.

Could you do it photographically? If you can photograph the head movements at a high enough frame rate, you should be able to calculate the acceleration that produced those movements.

It's about pressure and load, not movement. Also the sleds vibrate and shake enormously. Any method of measuring this has to be extremely precise, and also extremely responsive to changes over extremely short time frames to capture every vibration and shake in many different axes of freedom simultaneously. Before modern motion sensors I just don't think there was any way to do it.

Even if motion was enough, which I'm sure it isn't, film only captures a flat image from a fixed point, or the camera is moving itself so you have to compensate for that. Springs and wires suggested by others have way too much hysteresis to cope with the extremely rapid vibrations and changes in orientation.

When you are looking for something like 80g (as is the maximum of the graph), then the head would move something like 4 cm between two frames, which suggests to me, that you probably want to have a lot more than 100 fps, since you want to get a good handle on the highest accelerations.

On the other end of the scale, the positional error from the camera sensor is something like

  width of frame / horizontal resolution

which should give you something in the mm range, and it seems possible to get head tracking in the same order of resolution, which should limit your acceleration measurement for low acceleration.

Overall, probably doable but you are replacing a 30c accelerometer with a multi year research project.

The other thing is that with many sports there were massive "improvements" in tech, especially regarding speed, meaning that what was a "safe" sport until maybe 50-60 years ago is now putting human bodies to and over limits they have not yet evolved to endure.

And to top it all off, doping has also gone off the rails with no one having any idea about the long-term effects of the stuff athletes are taking - especially regarding cross effects between medications, shoddy manufacturing leading to contaminations and other such issues.

Except measuring G-forces acting on a helmet doesn't directly relate to the forces acting on the head within it.

Why not? To me (a layperson in the physics of helmets and brains), they would seem to be quite directly related, especially when the cause of the gees is not related to an impact.

Helmets are padded, deform, or have suspension systems to decelerate any impulse that the helmet takes. This should cause a very different g-force spike between what the shell of the helmet experiences, and what the wearer experiences.

If the helmet didn't decelerate any impulse to the head, the range of impacts that it could theoretically protect the wearer from would actually be quite narrow.

> If the helmet didn't decelerate any impulse to the head, the range of impacts that it could theoretically protect the wearer from would actually be quite narrow.

This is wonderful understatement: “If the g-forces between the helmet and the brain weren’t different, the helmet wasn’t a helmet at all!”

A helmet can just protect the skull So a helmet doesn't have to be padded or help with g-forces to be a helmet. I think helmets were initially just to prevent skull fracture.

well, yeah when those helmets were worn by knights in the middle ages, but many helmets nowadays have to withstand g-forces to do their job.

If English was like Italian there might be a specific word for a non g-force withstanding helmet and otherwise.

And every war in the first half of the 20th century.

Knight's helmets were actually heavily padded, with most knights wearing heavy padded linen garments over their torso, arms, and head. Nicer helmets would have a leather strap system as well. Many helmets appeared to have integral liners, although the linen has rotted away and only left the rivets where they were attached.

With the advent of plate armor, offensive techniques began to focus on concussive force in order to disable combatants, including war hammers and maces. A blow to the head with only a bit of steel between your skull and a mace would be a fatal experience without any padding. Even with padding it was notorious for deafening and disorienting knights, due to noise and (presumably) brain injury.

It's a lucky thing that Sir Ulrich Von Liechtenstein was such a cunning champion in both the sword and joust or he might have ended up with all the brains of poor Wat.

A helmet without padding can protect against low energy penetrating impacts; think BBs or similar. They do this by spreading the impact point out over the entire head; preventing penetration or fracture.

But for pure G forces, you’re right. A helmet with no padding is effectively similar to not having a helmet at all.

The helmet is padded.

Your head is in a suspension within the helmet, much like your brain is in a suspension within the skull.

Your skull can accelerate at extremely high G's without harming the brain as long as it stays well within the modulus of the suspension. i.e. not for too long. The helmet adds another suspension on top of this.

You are confusing the ability to sustain the g-force by the brain with the force experienced. A padding on the helmet helps in event of collision, not the g-force which is due to the acceleration

How is a collision not just another acceleration?

In a collision, the helmet gets hit and, subject to some degree of padding and such, transfers the impact to the head. If you’re sitting in a sled and the sled accelerated rapidly, your head and your helmet must accelerate with it, although you may not perfectly track its path. Unless the helmet is also secured to your torso, all the forces on the helmet come from your head. You would need tremendous acceleration of your head to meaningfully deform any but the softest padding.

Sure the helmet helps with g-forces. Say you're going down the track and enter a tight turn. The helmet contacts the side of the sled at say 5 g's because the helmet is hard and so is the side of the sled. The brain impacts the helmet, which impacts the side of the sled, but the padding in the helmet spreads that impact over a greater period of time, since the padding compresses, so the brain only experiences a g-force of 2.

The helmet is a light mass, which can move relative to the head, and experience drastically higher g-forces for very short times. Sampling an accelerometer at kHz rates as a helmet is vibrating (without translating all of the vibrational energy into the head), will measure larger peak G's than the head experiences.

For this reason, when studying head impact for football players, they mount the accelerometers in the mouth guard. This more accurately reflects the real g-forces, as the teeth (ideally) don't move much relative to the brain during an impact, and have the appropriate reference frame.

The differences can be extremely large. The peaks reflected in this graph are absolutely not real and should not be taken literally.

What would you say is the maximum frequency that the helmet is coupled to the head at? Filtering is a trivial operation and I would expect the g forces felt by sledders to be sustained for the better part of a second. You lose a lot of fidelity, but if you’re still over the line then you know there’s a problem.

According to the article, the tracks are designed for sustained g-forces of 5Gs. These are the accelerations that are sustained for greater than a second. According to the annotations on the NYT's plot, the g forces which are most alarming are from when the helmet makes contact with the ice. This is precisely when the helmet accelerations would be expected to differ wildly from what the head is actually doing.

A low pass filter could easily be used to remove the impacts with the ice, but minus the impacts with the ice there's nothing in that acceleration trace that should be alarming for a professional athlete. The high frequency content is the concern here.

Yes, but notice that the shock goes sled -> spine -> head -> helmet in sledding and other player -> helmet -> head in football. So, strapping an accelerometer onto the helmet looks quite a bit better for sledding. (And especially for a first measurement.)

Yep, except in the cases where the head hits the sled or ice, as mentioned in the article. I suspect these are responsible for large (and artifactual from the perspective of the brain) transient spikes in measured G-forces

The whole point of a helmet is to absorb impacts. It's designed to dampen rapid accelerations so your head doesn't have to.

Do you think there's a material inside the helmet that can stop gravity?

Are you perhaps confusing impact with g-force? A good analogy would be a fighter pilot or an astronaut. The g-force is felt all over the body irrespective of whether you are wearing a suit or not.

The g-force spikes in the graph are due to impacts. It even says that on the graph.

It doesn't matter. You don't have to measure something perfectly and exactly to decide if it's an avenue of research worth exploring. If turns are designed to limit G-forces to 5G, and you put sensors on your helmet and they're reading 55G when your head hits the ice, that's probably an indication that further research is required.

You could have >55G when you stay in your room and bump your head against the wall

Yes, but you don't do that over and over again every day. Unless you like practicing bumping your head against a wall, but it's not a sport I've ever heard of.

ok - You have a point. Researching this area is worth it. Can you see that the others also have a point? That is, measuring helmet g-forces is a proxy to find head g-forces but they are not one and the same?

However, the real issue is not g-forces on _corners_ it is the g-forces on _head/helmet hitting the ice_ where 5 g from the first are rounding errors compared to 55g from the second.

"measuring helmet g-forces is a proxy to find head g-forces but they are not one and the same"

I don't think I said they were, just that it's a worthwhile thing to measure to see if you should start finding a way to get better measurements. I agree they are not the same and this is not a definitive study. This is often how science works, you do something quick and easy. If it came out that it was only ever minor forces on the outside of the helmet, it's probably not worth investigating since the forces inside the helmet will be even less. If they're huge, repeated forces, it's probably worth trying to come up with a better rig to measure actual force and see if it's above a healthy level.

I once went on a bobsled ride on an old Olympic track. 4-person bobsled, two professionals controlling the sled, two passengers. I think we didn't go into the final turn properly, and the sled rode high and the brakeman (rear) hit his helmet on the side of the track. He was a bit dizzy afterwards, but otherwise it wasn't treated as a big deal. If this kind of thing just happens on relaxed tourist runs, I can only imagine what's normalized in training and competition.

Bobsledding isn't as popular as the majority of "international sports", putting it into that category doesn't justify its strong examination. I'm not surprised personally, it's a niche sport without human-on-human violent contact; something like American football is more perceptibly dangerous than some of these rare Olympic sports.

That bobsleigh track was designed specifically to be faster than any other. They had to close part of it for the actual Olympics because someone died in training right before hand. They went for records instead of safety.

Reply on news.ycombinator.com