Hacker News Reader

Comment by njarboe | Hacker News Reader

With automated driving systems you can put the vehicles really close together with high speed and get throughput equal to train systems with single passenger cars. These same tunnels can also have vehicles in them with higher density seating (call them buses, if you like). If you put these vehicles close together you get a train. With this system you can have both cars, buses and trains running all in the same tunnel. Charge each by how much space they take up and vary the price so there is never a backup to get on. Seems like a transportation dream come true. Everybody gets the system they want to use.

View on HN

That's not solving the issue: single-passenger cars are just extremely inefficient passenger density. Technology might allow you to pack them to 2× or 3× density, but the space efficiency of a single railcar is closer to 100× that of a single-passenger car.

How close can you put trains together? For BART it is 24 trains max per hour in the tube and with a new switching and control system they claim they will get up to 30[1]. They go about 60mph. So two minutes between trains. Think about how much space that is: 2 miles between trains. Trains have to carry 100x more people otherwise they would not work at all. Imagine a freeway where cars had to be two miles apart at max density and only going 60mph. You would never see another car while you were driving.

The plan would to be able to pack cars at 100x the density of trains.

[1] https://www.bart.gov/news/articles/2016/news20160502-0

If your signalling and switches are good enough, you can get to about ~30s headways. You can't really go faster than that because station dwell times will screw you. Scheduling this theoretical maximum is a daft idea, though, because any mishaps, and you have no spare time for recovery. Metro systems in practice usually end up being scheduled around 90-150s headway, so about 24-45tph.

There's one catch for trains though: most trains aren't dispatching single cars. Rather, trains on busy systems are generally 8 or 10 cars that move as a consist. If you factor that in to your example, you don't have 120s between cars, but rather 12s. That might sound like a lot, but you should have 3s between cars for safe driving anyways, and train cars carry well over 4× the number of people in individual cars. Even if every road car were full Jitney vans.

Apart from tolling based on footprint, that sounds exactly like what we have now. Automation might increase the capacity, but individual cars will never be as efficient the public transit we already know how to build. You're never going to get the cars closer together than people on a train.

Did you read what I wrote? You can put buses into this system and charge by how much space the vehicle takes up. If a car is 1/2 the space of a bus then the bus pays 2 times the price and the people in the bus pay 1/20th the price. If demand is high enough then car travel will get very expensive and the system transitions very nicely to a train system with only large, packed buses traveling in it.

This system is also nice because people could choose all types of different ways they would like to travel. From sleeper cars, to mobile offices, to cramped seats on a bus, to stand-only packed in like-sardines, we are now in a Japanese subway situation.

Or you could just make it a damn subway system at the begining and be done with it. This seems like a damned inefficient and expensive way to get a subway line.

Even worse, it sounds just like an underground highway with streetcars and trams.

But as it runs on skates, do they want everybody to input their routes first? And clearly, it's going to cost so much now than surface highways, but economics is lost here, for people are convinced that a maglev train in a pneumatic tube will cost 6 billion, but a high speed rail will cost 80 billion. The name Elon musk is like magic to dispel any economic sense.

Or you could just put a train in the space taken up by 10 automobiles. In LA, each rail car can hold roughly 100 people at capacity, and each train has 3 cars, so we're looking at a 5x-29x increase in capacity over Musk's underground freeway proposal.

This comment would make only sense if the train density on the track we're the same as the car density in a tunnel.

Conceivably, the utilization of the tunnel could be way higher than the utilization of train tracks (most of the time: empty)

Trains can run every 3 minutes, that means the throughput of passengers is roughly 900 passengers every 9 minutes.

You're not getting 150+ cars in that tunnel during the same time period.

Doesn't look that way in the bay area on the caltrain line from San Jose to SF. Tere, due to limitation and NIBYism along the tracks, but part of the argument of the boring folks is that a tunnel would not face the same political headwind that a decent train system would require.

No you can't, you need to leave stopping distance for safety reasons. If anything AVs will leave more room between cars than human drivers.

Citation needed?

Like, the drivers manual? We're talking about basic driving skills here. Cars need room to stop, the higher the speed, the more room they need to stop safely. Most human drivers aren't perfect and leave a lot less room than they should, AVs won't share that flaw.

AVs can react faster than humans - if all the vehicles in the tunnel are controlled automatically, distances could conceivably shortened quite safely. Not sure why "citation needed" was downvoted, it's far from clear no matter where you stand on the issue.

Reply on news.ycombinator.com