I'm not sure what you're purporting to show other than two people disagreeing with each other, without providing evidence... Sure, an individual subway car has "more capacity," but does a subway system have more capacity? For example, I rarely see BART trains come within five minutes of each other. A tunnel that allows cars within 10 seconds of each other could provide more capacity than a tunnel with higher-density trains spaced further apart. Average density across the tunnel system could be higher (aka more throughout) with autonomous vehicles even if individual transport units have lower density (SUVs instead of train cars).
Off the top of my head, here are issues train cars face with managing density that autonomous vehicles wouldn't face:
1. Needing to stop at stations where few people are getting on or off, meaning that trains behind you must be at least (MAX_SPEED * STOP_TIME) behind you. With cheap tunnel boring + AVs, you can quickly and cheaply build tunnel branches that skip exits, so individual cars need less space between them (you branch off to exit; no need for extra rails or switching technology, since it's built into the cars).
2. Entrance/exit from the train car: with trains, any jerk can run in as the doors are closing, preventing the train from leaving, backing up the trains behind you and leaving a huge gap with trains in front of you. With AVs, you're already in the car.
3. Trains have fairly long stopping distances, increasing the distance needed between trains for safe operation. Some quick Googling indicated that an 8-car passenger train traveling at 80mph can take over a mile to come to a complete stop — it's hard to imagine a car with such long braking distances. That puts a pretty hard lower bound on how densely you can pack an ordinary subway tunnel.
Maybe I'm totally wrong, but it would be more useful to give actual data and evidence than knee-jerk shutdowns. It's not obvious to me, a non-train-engineer (like many on this site!), that the dismissive tweet is particularly meaningful.
The equivalent of BART in Paris (RER) has 90 seconds intervals between trains in peak time - and thats the regional train system, not the metro. It carries 50k pax per hour in each direction - that would be a lot of cars, like eleven cars a second at 1.2 person occupancy.
You mention "cheap tunnel boring" - maybe that is the keystone of Musk's plan that will change the paradigm... With a little knowledge of tunnel boring I fail to see how to disrupt its crazy complexity, especially in urban environments whose density extends underground already: in my neighborhood (La Défense business district) a whole new station is being build wholly under existing construction and interconnected to existing subterranean works, without any disruption - insanely good engineering and insanely expensive.
One of Elon Musk's ideas is that underground you can have a three dimensional system of tunnels and that this can help solve some of the capacity problems.
As dense and as cheap as Musk's technological vision becomes, a question remains: why carry a car at all instead of carrynig people ? Why carry mostly dead weight ? The denser the network, the more acute that question - if Musk's vision is an ubiquitous three-dimensional underground transport mesh, then why does it seem to conscientiously avoid being a public transportation system ? Does Musk believe that saturating the city with autonomous vehicles that use tunnels as network backbone can be more efficient than heavy public transportation networks ?
Peak headtimes on BART are a train every 2-3 minutes, with trains carrying up to 1400 people (let's call it 1000 for the sake of argument). So at the low end, that's around 20,000 people an hour. If you had cars with a headway of 5 seconds then that's 1200 cars an hour, so you'd need an average occupancy of over 16 people per car. That's unrealistic, so yes, a subway tunnel has more capacity.
That's assuming there's only one tunnel. As the article mentions the point is to dramatically reduce the cost to bore tunnels, so you can bore more of them.
If you could bore 4x the tunnels, you'd need a capacity of 4 people per car to match peak BART. And... hey, that describes pretty much every car in existence. But using the BART comparison: recent estimates [1] put each additional mile of BART track at over $1 billion. So if it really costs $10MM/mile for the Boring Co, you could bore 32x the tunnels and beat BART capacity by a factor of two assuming even single occupancy vehicles, at about a third of the cost.
cf London Underground - Central Line does ~260M journeys a year which is ~40k/hour assuming an 18 hour day. (I suspect peak times will be ~50% higher than that average.)
Even the Waterloo & City Line, the most unloved of all lines, manages 3.5k/hr over 260 days (16M/year, 5 day week).
> Sure, an individual subway car has "more capacity," but does a subway system have more capacity? For example, I rarely see BART trains come within five minutes of each other.
The BART is a particularly poor example. There are _tram_ systems (ie partially-segregated rail) which do better than once every five minutes.
Point to point transport in most cities is not at all a solved problem.
While this was demonstrated with a SUV there’s no reason why this can’t be used for a larger van carrying 6-12 people and tied to a carpool service. 12 people hurtling down this each second is pretty decent capacity, especially multiplied by many tunnels that do an even better job at getting people to the last mile.
>There is no universe where an SUV holding 4 people is more space efficient than a Skytrain car.
Except universes where a train doesn't pick you up at your house, drive you to the front door at work, take you to your kid's school after work, then take you to drop the kid off at karate practice, then take you directly to the grocery, then back to karate practice, then home but allowing you rapidly move your vehicle through a congested city thereby reducing surface congestion making overall travel conditions more efficient and reducing the need for car parks at public transportation stop for people that live in suburbs.
75 seconds is really, really long. Cars on a highway don't leave 75 seconds in between each other; if a cutting-edge train does, that starts to imply that density issues do exist for subway systems.
And how expensive was it to build the train? If you can build 3x the tunnels for AVs as you can train tunnels and trains for the same price, actual throughput dollar-for-dollar in infra spend would be much better with Boring Co tunnels.
Update with data: it cost $2 billion to build 50 miles of Skytrain. Apparently this prototype Boring Co tunnel cost $10 million per mile. That's a 4x cost advantage for Boring Co mile-for-mile, and I imagine they'll try to strengthen that advantage even further; after all, that's their whole company's purpose: cheap, efficient boring.
Right, but 15 SUVs 5s apart carry 105 people, and if they can actually travel at an average 150mph, that's faster than the average speed of Skytrain's fastest line by another 3x — so, now we're looking at 315 people moved through the space at the same time.
Factor in cost advantages of the tunnel — 4x cheaper mile-for-mile for the prototype, it seems like? — and the same dollars spent on Boring Co tunnels could actually move a lot more people than Skytrain: you can build 4x the tunnels, moving nearly 1300 people instead of 500.
Even if SUVs can only match the fastest average Skytrain line (50mph), that's still fairly competitive even at the current cost per mile of a Boring Co tunnel. Assuming they can bring cost down further (which is the goal of the company), they could still end up beating trains in terms of people moved per dollar spent.
Listen, I'm not saying it'll happen. But the dismissiveness here reminds me of the "Model 3 will never cost less to build than the price to sell it!!!!!" or "Tesla will never make more than 1000 cars per week!!!" or cash crunch or etc hysteria. I guess Elon attracts that by constantly missing deadlines, but he often still delivers; he just delivers a year or so late.
Despite your pretend block quote, that isn't a direct quote from me (you're taking quotes from someone else in the thread and splicing them into my post, using your made-up version as a strawman). I said 5s apart, which is over 1000ft of spacing — nearly a quarter of a mile, plenty of room for braking — at 150mph; the 15 SUVs is because 75/5 = 15, and we're comparing it to a train system with mandatory minimum 75s gaps.
This plan is like when a city widens a highway but doesn't touch other parts of the road network? What happens? It just moves the choke point somewhere else in the network.
Agreed 1s gaps are too tiny — seems unsafe and that'd be a lot of elevator traffic. 5s seems doable though: assuming 5s gaps, you need an elevator system capable of moving one car every 5s, and then the tunnel is optimally saturated. That doesn't seem insurmountably hard: two elevators that take 10s each to go from surface to tunnel would be enough, along with a bit of on-ramp tunnel to allow cars to accelerate to speed before merging. Throw in a third elevator or a fourth for redundancy while you're at it because elevators are way cheaper than a mile of tunnel, and seems doable.
You're assuming you can get cars into the elevators at 5s intervals.
Almost all assumtions in the discussion start directly in the tunnel, and sometimes at the elevator. Almost no one thinks about: how will cars get into that thing from the road, how the cars will get out of that thing onto the road.
It's probably just me but I really wouldn't get into anything that was going to be travelling 150mph at 1s intervals. All you need is a tire to go (or something loose on the floor or a dead battery or ...) and you've got not only a huge twisted lump of metal and body parts to untangle, you've almost certainly ruined your tunnel.
You're comparing a prototype with a real life operating system that was built to be um, you know safe and good, unlike what has been shown with this prototype.
The fact is that a 500 capacity train running at 75-100s headways destroys this system capacitywise.
Musk has a long, long way to go before he shows that this idea of his is better than the 1986 era technology at the core of Vancouver's system.
> The fact is that a 500 capacity train running at 75-100s headways destroys this system capacitywise.
This is like that linked tweet: all assertion, no data. What are you using to estimate the theoretical throughput of a Boring Co tunnel system vs Skytrain? Let's debate that. Right now it's all just thin air.
Assumptions I'm willing to believe, and thus make me interested in this prototype's viability:
1. Cars need less space in between them. 5-10s gaps seem more than reasonable based on highway driving. This seems especially true since these are operated autonomously in highly controlled environments, meaning we might need even less distance between cars than free-wheeling, human-operated highway driving.
2. Most train stops are useless for most people — maybe 3x stops per ride on average? Having only one stop per ride would dramatically improve average speed, since a smaller percentage would be spent slowing down, speeding up, or waiting at a station.
3. Smaller tunnels are cheaper to dig, meaning dollar-for-dollar you can build more of them to make up any remaining capacity difference.
Off the top of my head, here are issues train cars face with managing density that autonomous vehicles wouldn't face:
1. Needing to stop at stations where few people are getting on or off, meaning that trains behind you must be at least (MAX_SPEED * STOP_TIME) behind you. With cheap tunnel boring + AVs, you can quickly and cheaply build tunnel branches that skip exits, so individual cars need less space between them (you branch off to exit; no need for extra rails or switching technology, since it's built into the cars).
2. Entrance/exit from the train car: with trains, any jerk can run in as the doors are closing, preventing the train from leaving, backing up the trains behind you and leaving a huge gap with trains in front of you. With AVs, you're already in the car.
3. Trains have fairly long stopping distances, increasing the distance needed between trains for safe operation. Some quick Googling indicated that an 8-car passenger train traveling at 80mph can take over a mile to come to a complete stop — it's hard to imagine a car with such long braking distances. That puts a pretty hard lower bound on how densely you can pack an ordinary subway tunnel.
Maybe I'm totally wrong, but it would be more useful to give actual data and evidence than knee-jerk shutdowns. It's not obvious to me, a non-train-engineer (like many on this site!), that the dismissive tweet is particularly meaningful.