Sunday, 10 September 2017

Stopping distance experiments part 1

The Charlie Allinston case has not only raised the issue that fixed wheel, chain-only-braking bikes seem unsuited for the urban environment, but that cyclists themselves are criminals.

This was shown by a Met office video demonstrating how a mountain bike could stop in three metres, while a fixed wheel bike took 12 metres. Which, from the public video, looks pretty dubious.

Ignoring the thinking distance aspect of the problem (to be covered another time), neither of those bikes appear to be doing 18 mph. Martin Porter asserts that with a maximum deceleration force of a bike of 0.3g before the rider goes over the front bars, any bike is slower to decelerate than a car (max force 0.5g). We don't know the exact values. What we do know from experience is
  • It's the front brake which does most of the conversion of kinetic energy into heat
  • As you brake, you go forwards
  • The back wheel, without weight, goes into a skid
  • And it always seems to happen faster on tyres with less contact area (narrow tyres, knobbly mountain bike tyres on tarmac, ...)
Mountain bikers will know that to avoid going over the bars on a hard brake or steep descent means sticking your bum out the back, which is partly why dropper posts are so popular: lower C of G, better control on the descent.
Returning to the Allinston case, the police stated that Charlie had 6.5m to stop, and that because a mountain bike could have stopped in 3, his removal of the front brake made him culpable. But: they haven't show the actual CCTV of of the collision, so we don't know exactly what happened
What we can do though, is do the experiment proposed by the People's Cycling Front of South Gloucestershire: try to stop from 18 mph in under two car lengths.

Here then is our summary:
  1. an experienced road and mountain biker cannot stop a road bike with touring tyres in the distance of two parked cars and a 1m gap between them. 
  2. You can bring the speed down to about 6.5-7.5 mph (update: see the bottom of the post)
Equipment: a team road bike, "roadkill", purchased for $800 in 2000 somewhere in Oregon, a US state which is now charging people a tax for doing so.



  • Tyres: Continental Top Contact 700x28 touring tyres, tyres which focus on all weather control and braking over rolling resistance and speed. That is: their braking should be as good as you can expect from any road tyre of a similar diameter.
  • brakes: front and rear Shimano 105 rim brakes (from 2000), cables redone 24 months ago, pads Shimano 105/Dura-ace/Ultegra inserts
  • Wheels: Mavic rims, hope hubs, again 24 months old and no rim wear.
  • Garmin bike computer to determine speed from the GPS constellation and display speed as part of preparation for braking
Overall then: the brakes and tyres aren't going to degrade performance compared to any other road bikes with rim brakes.

Experiment:
  1. Flat, traffic free road with enough visibility of the southern sky that GPS signal will work (Trivia: US Navstar satellites never orbit  > 54 degrees north or south, so above the lake district getting a signal is harder; for Galileo details ask one of our local engineers)
  2. Two family cars, a "normal" gap between them (nobody had problems fitting today).
  3. Turn on the Garmin to record speeds.
  4. From a distance down the road, bring bike up to 18 mph & then coast briefly.
  5. As you reach the front of the first car, brake as hard as you can safely, arse out the back and down as learned over the years on the MTB.
  6. At end of the two car lengths, see what your exit speed is.
  7. Repeat a few times.
Not the most rigorous, but with speed numbers coming from 31 orbiting atomic clocks it's as good on the flat as anything else.

Results



  1. Getting to 18 mph on the flat does actually require effort, if attempted over a short distance. (This may make cyclists reluctant to shed that speed)
  2. Even with warning and planning, you can't stop a bike in 2.5 car lengths from 18 mph
  3. You can get down to 6.5-7.5 mph
Conclusions
  1. If the met police video released to the media was the one used in court, then the qualify of the experiment has to be contested.
  2. As well as the choice of the reference "not a fixie bike" as a mountain/hybrid bike with what appears to be wide surface area road tyres, its not clear that they are doing 18 mph when they get to the marker points where deceleration is meant to commence.
  3. If someone else can stop from 18 mph to 0 mph in 3 metres, we'd love to see it.
  4. Given warning, you can get to 6-7 mph, which may lead to reduce risk/scope of injury.
  5. Given that cycle/pedestrian collisions at what for on road speeds are "low", we shouldn't be doing any cycle-paint-on-pavement bike paths, as they are engineering in danger
It'd be nice to see the logs of the police instrumentation data from their experiments. In fact, its something a defence lawyer should have been demanding: "show us the hard data"

Someone should run Bristol Bike Week event "can you stop in 6.5 metres with and without advance warning", to see what other people can do.

2017-09-11 Update : there's a flaw in the experiment: the measured exit speed is inevitably going to overestimate the actual one on any attempt to decellerate in a few metres.

Velocity (Speed) is distance/time: (d1 - d0) / t
But: what is the sampling interval of bike speedos (GPS and on-wheel?).
  1. Not clear from GPS (though looking at the GPX file would probably show it), but 
  2. on a 700x28 wheel the circumference is ~2.1 metres. 
  3. Therefore the magnet on the wheel will only send packet to the bike computer every 2.1m of travel. 
  4. Therefore the minimum distance which can be used to measure velocity is 2.1m. 
  5. 18 mph is ~29 kmh
  6. which is ~8 m/s
  7. If you are trying to come to a halt  from 8 m/s n 6-8 metres then that's only four revolutions of the wheel, so every revolution will have to include a lot of deceleration: 
  8. Assuming constant deceleration you are going have  to enter that final wheel rotation in travelling 1/4 of your entry velocity
  9. so: it's inevitable that the distance travelled in that final rotation is going to be "something" between the entry velocity and the exit one
  10. which is what the bike computer will end up displaying: it will overestimate the actual value
All we can really say is "the bike was still moving 10 m after attempting to decelerate from 8 m/s to 0 m/s"

How to do it better? Well:
  1. you could let go of the brakes after crossing your chosen endpoint, coast for a few wheel lengths and so give the computer a constant velocity to measure. 
  2. with the sub-cm resolution promised by Galileo's premium channels and a receiver set to sample many times a second, you could build a fully accurate model of decleration
  3. you do rigorously measure the total distance travelled, assume constant deceleration over the distance and then work backwards from there to infer your velocity at the 6.5 metre mark
Anyway, it's mostly moot due to the thinking time even before you reach for the brake levers. Followup on that in the week, this time with video data. Essentially: you don't react fast enough.

1 comment:

Neil Hobbs said...

Interesting read. Yes the rider was a bit of a knob but he shouldn't go down for this. This could happen to anyone, pedestrian checking their facebook status walks in front of you but you are the one in the wrong?
Makes you wonder why his solicitor has not challenged "the facts"