5 years ago you probably would have said "we can't have a self driving car... there is <problem1> that we can't surmount." Yet, now we are on the brink of it. Hell, even at the start of this year I would have laughed if you said we will have aerial drones delivering amazon goods. Yet, here we are with a pretty impressive POC from amazon. we can merge onto a freeway why can't we merge these hyperloop pods so we can have many stops?
Because deceleration takes time, unless you want to be flattened to the front of the vehicle. The more stops you have, the lower the overall speed will be at which this thing operates. High-speed trains in Germany that can drive up to 350 km/h often only average around 120 km/h over their total distance because with stops in between you spend a lot of time accelerating and decelerating. And where stops are more frequent you can't even reach the full speed since you have to start breaking already before you're done accelerating.
There's nothing inherent in physics that prevents self-driving cars, but unless you want to jump out of a moving hyperloop pod, it has to come to a stop and accelerate again; and it does so slowly enough not to kill its passengers.
I think we can tolerate about 0.5g of deceleration, which shifts the direction of gravity by 30 degrees. So, about 5ms⁻². To get to interesting speeds such as 500ms⁻¹ (about mach 1.5), it would take 100 seconds, and 25,000m (25km, or about 15 miles). Assuming maximum acceleration and deceleration, it would take 3 minutes and 20 seconds to travel over 50km. At first approximation, separating stations by 50km looks doable.
But that's not the whole story. If we have intermediate stations at all, we need to ensure all cars have the same speed when they get to share the same main tube. We need acceleration tubes the same way we need acceleration lanes in regular roads. (How to plug those tubes to the main one is left as an exercise to whoever builds this.) Anyway, those tubes need to be 25km long.
But that doesn't mean they have to be separated by that much. You could have a station every 5km if you're willing to run 5 acceleration lanes in parallel (and pay for them).
This gets even easier once you get close to the destination: even in the main tube, cars need to decelerate, starting at 25km from their destination. 10km away for instance, the speed is already down to a little over 300ms⁻¹. To join the lane at that point, you need only 10km to accelerate. Likewise, if you need to join 5km before the destination, you need to accelerate for 5km as well.
In other words, you could very well have stations as close as 5, 10, or 20km from the urban centre. You won't go very fast, but remember that at those speeds and accelerations, it's going to take less than 5 minutes anyway.
---
If trains didn't have to stop at every station, and used acceleration and deceleration rail-roads where appropriate, their average speed would be much higher. First, they can't for 2 reasons: rail-roads are expensive and take land (so does an Hyperloop tube). Second, passengers need to go in and out. This means many stops for the same train. Hyperloop's uses individual, so it doesn't have that problem.
> I think we can tolerate about 0.5g of deceleration
Untrained human beings can handle minutes of linear deceleration (perpendicular to the spine) at 12 g, and at least a second at 25 g. Strapped, mind you. And the limits are higher in backwards-facing seats (eyeballs-in is ~17 g untrained) with the additional advantage that very little strapping is needed.
to be fair the test that showed a lower bound on survive-ability caused the fellow eye damage. (wikipedia says it was measured at a peak "eyeball out" g-force of 46.2g and 25g for 1.1 seconds.)
I'd say 1-2g over a span of a few min might be tolerable by the general population but might not be need to be researched for suitability for children, the elderly, and the disabled.
There are a lot of strategies for handling this. One of which is, accelerate the pods on a separate, shorter (semi-circular?) track before merging into the main "pod stream". Similarly, decelerate on a separate, shorter (semi-circular?) track. There was some prototype (or theory) of this discussed in a link on HN several months ago in which a Chinese train station accelerated (or decelerated, I forget) a car to match the speed of a passing train to append that car to the train as it went by.
Just because you need to accelerate and decelerate doesn't mean you need to do it on the same shared track as the main line.
In the same spirit, small pods allow far more flexibility than a bus. With small enough pods, i stop at my destination only, rather than every destination that anyone needs. I'd guess 4 or 6 is a useful size, about the same as cars.
4 years ago I was telling people we would have self driving cars in 5 years. So I'm very pleased with the progress.
It remains to be seen how effective pumping can be, but as envisioned hyperloop is a point to point system. Switches are very difficult and must be done at low speed a) because of the slipstream disruption and b) because branching greatly increases the interior volume of an enclosed tubeset. It can be done.. but it's a level above the basic infrastructure.
