Hacker News Reader

Comment by kumarski | Hacker News Reader

Something I wrote a while ago.... w/ back of the napkin calculations.......

Why I wouldn’t invest a money into Hyperloop

- Stabilizing a single fault line risk pylon is more than $250K.

- How many million are needed for vacuum pumps to evacuate 100+ million cubic feet of of pipe to 100 Pa?

- Hot air discharge needs to go somewhere. For every 1 bar pressure, you need ~200 to ~400 cubic meters of volume which is larger

- This seems very much like one of those Andy Grove Fallacies.

- The hyperloop is a mega engineering project on the ground. Nobody on their team is a civil engineer. Looking at their team objectively, there seems to be a mismatch of competency.

- At its core, the science i good, the cost-economics do not work. Das ist nicht gute.

View on HN

I don't think it's right to use figures that can change to disprove an Elon Musk concept. If you did that back in the day, you would say there's no way Tesla could work because the cost of batteries is prohibitive and the supply too limited.

This is the real insight.

Yes, but we could see that batteries needed to get X better in order for us to achieve Y results in the future. But how much better do vacuum chamber technologies have to get before you can make the biggest one ever built, that also has a high-speed capsule traveling through it, that is also impervious to armor piercing bullets? Furthermore, imagine if every airplane had to fly the exactly same path, and if any failure on that path prevented (at the very least) all subsequent planes from traveling that path. And finally, how are security lines going to get shorter if the stakes will be at least as high as airplanes?

I don't know the answers to these questions, but what I do know is:

1. Elon Musk has thought about them more than we have.

2. If his initial concept turns out to be incorrect in some way, he will change it until it is workable.

Are airliners impervious to armor piercing bullets? As I understand it, a small bullet penetrating the loop will cause a minor release, not a catastrophic event. It's not correct to hold the Hyperloop idea to standards that we don't apply to other forms of travel.

There's a very pertinent historical reference point: Ferdinand de Lesseps. He built the Suez Canal from 1859 to 1869, a task which was considered by many to be impossible or at least impracticable [1]. After achieving resounding success, he then turned to Panama to try building the Panama Canal. In this venture, he turned out to have several very wrong ideas (it's not feasible to build a sea-level canal in Panama), and the French effort failed so hard it actually brought down the French government [2]. He was successful in the first venture but for different reasons than initially assumed (namely, the Suez Canal isn't all that challenging from an engineering perspective; it's just capital-intensive), and his failure in the second venture comes from misunderstanding the problems.

[1] There had been canals in the past that traveled from the Red Sea to the Nile delta, but these invariably silted up rather quickly (the Nile was sediment-heavy, and wind-blown sand is quite common in the region). Thus there was historical evidence favoring the idea that the dredging maintenance fees would exceed the revenue from passing ships.

[2] Of course, this was the Third Republic, which is notorious for having provided series of weak, unstable governments replete with scandals that prompt new governments. Ironically, the Third Republic was the longest-lasting form of government France has had since the Ancien Régime.

To this day, the Third Republic is still the longest lasting.

Assuming your are correct and Branson knows it because he hired good engineering advisers. Why would Branson get involved ?

- It is a cool project he wants to be part of and he is willing to pay to see. (I would probably do that)

- The benefit is not to reach the goal but all the technology that will be invented trying to reach the goal.

> Why would Branson get involved ?

Branson is notorious for burning other peoples’ money. Look at Virgin Galactic: it’s New Mexicans’ [1] and Floridians’ [2] tax money plus some duped Emiratis [3]. (They also pre-sold tickets [4].)

[1] https://www.theguardian.com/us-news/2015/oct/04/space-travel...

[2] http://www.airnewstimes.co.uk/space-florida-promises-virgin-...

[3] https://dealbook.nytimes.com/2009/07/28/uaes-aabar-takes-sta...

[4] https://www.smithsonianmag.com/smart-news/

Branson's business model is selling the right to use the Virgin brand to other companies. The article doesn't say he made an investment. It is rather more likely that they paid him.

So basically Trump with more class?

...and fewer bankruptcies.

Although if you believe Private Eye about the state of Virgin's finances (convoluted, shifty, and blistering opaque), this may or may not be true.

Is that actually the case? My impression is that the vast majority of companies bearing the Virgin brand was either started by Virgin Group, bought by them, or has seen a significant investment. Many have since been divested again, and continued to license the brand, but I've not heard about any cases where they just licensed the brand without other financial involvement. Maybe I've just not paid attention?

https://en.wikipedia.org/wiki/Virgin_Group#Subsidiaries_and_... shows the majority being 100% owned by the Virgin Group, though notably none of the airlines nor rail operators are.

> The article doesn't say he made an investment.

Actually, from the article:

> Hyperloop One has received a significant investment in Hyperloop One

And from the page linked to in that paragraph:

> Delighted to announce Virgin Group’s investment in Hyperloop One

Not exactly ambiguous.

When Musk first said that an LA<->SF Hyperloop could be built for 1/8th the cost of the equivalent HSR, I chuckled and then waited for the punchline, which still hasn’t arrived.

It’s a cool idea and I do want to see more prototypes and feasibility studies, but people need to get off the hype train (no pun intended) and be realistic about the cost.

> When Musk first said that an LA<->SF Hyperloop could be built for 1/8th the cost of the equivalent HSR, I chuckled and then waited for the punchline, which still hasn’t arrived.

Yeah, and that proposal was for something with much lower throughput, that didn't have stops along the way, and didn't even reach the same areas (they saved money by stopping outside both cities, which the rail line wouldn't do). It also assumed that the tubes could just be placed along the highway meridians (a lot of the money it was supposed to save was from this).

To be fair, if Elon worked in the project for 40 - 80 hours a week, he probably could get a team to accomplish it at 1/8th the cost or lower.

Consider he did this with rockets, getting them to under 1/10th the cost

He hasn’t done that with rockets. He wants to eventually, but he hasn’t yet. Let’s not let hype cloud facts.

He's getting very close, if one is considering domestic rockets. The standard-bearers were the Delta IV and Atlas V, which price out at around $17,000/kg - $20,000/kg to LEO, respectively. The Falcon 9 comes in at about $2,800/kg to LEO.

Comparisons with non-domestic rockets are complicated by state subsidies and differentials in labor costs. But the Falcon 9 is approaching the 10X mark even before reusability has been priced in. Both the Falcon Heavy (in a few months) and the (eventual) reusable pricing should overshoot the 10X mark substantially.

Where are you pulling those numbers from? On ULA's website [1] they say you can launch an Atlas V for less than half the $/kg you listed

[1] I went to rocketbuilder.com and both LEO configurations I tried were around $8,000/kg (and I didn't count the 20% deduction for "ULA added value" since the numbers there are debatable)

ULA has abruptly gotten way more competitive in their pricing. A few years ago they were not. Their "Best deal" for the Atlas V was a 36-rocket "bulk buy" priced the lowest-tier Atlas V 401 at $164M/launch[1]. That does 9,797kg to LEO[2], or $16,740/kg.

The GAO cites $164M/launch as the price for the cheapest Delta V[3] (is there some kind of threshold at $165M?), which when the Falcon 9 was introduced could put 8,500kg to LEO[4], or $19,294/kg. (It has since been uprated and gotten slightly cheaper).

Must've screwed up my calculations earlier, because those aren't tallying. Anyhow, as you can see, SpaceX is already around 7x cheaper compared to where ULA was, and should cross the 10x threshold shortly. Nice to see ULA responding to competitive pressure, however!

1: https://web.archive.org/web/20160324172526/http://www.ulalau...

2: https://en.wikipedia.org/wiki/Atlas_V

3: http://www.gao.gov/products/GAO-17-609

4: https://en.wikipedia.org/wiki/Delta_IV#Delta_IV_Small

For the "block buy", that price includes building the rocket but not the launch.

With rockets he was competing with fat and lazy government contractors for the most part. The hyperloop will be competing with already cutthroat airlines, the goddamn personal automobile, and, ...well, a fat and lazy passenger train company.

IMHO he'll be well over 1/10 the cost before he finishes the thousands of land deals he would need for this, even if he somehow got permission to build it over the highway.

> even if he somehow got permission to build it over the highway.

Hence the boring company.

Which won't do anything, because you can't magically make tunnelling cheap

You can't do it magically, but you can do it by improving the technology - see for example SpaceX.

SpaceX is cheap due to reuse and commercial of the shelf development compared to mostly military-government development.

Tunneling gets cheaper due to <...>? I would be interested.

I guess the Swiss would be interested too.

Musk talked about it here: https://medium.com/@giftedproducts/elon-musks-ted-talk-2017-...

Salient points:

"If you just do two things you can get to approximately an order of magnitude improvement, and then go beyond that. The first thing to do is cut the tunnel diameter by 2 times or more. A single road tunnel has to be 26 to 28 feet in diameter to allow for crashes and emergency vehicles and sufficient ventilation for combustion engine cars.

But if you shrunk it to 12 feet, what we're attempting, which is plenty for a skate to get through, you drop the diameter by a factor of two and the cross-sectional area by a factor of 4. The tunnel costs scales with the cross-sectional area. That's roughly half an order of magnitude improvement right there.

Then, tunneling machines tunnel half time and then stop and the rest of the time is reinforcements for the tunnel walls. If you design the machine to do continuous tunneling and reinforcing, that will give you a factor of two improvement. Combine that, and it's a factor of 8. Also these machines are far from being at their power or thermal limit. I think you can get a factor of two, maybe even four or five on top of that.

There’s a fairly straightforward series of steps to get somewhere in excess of an order of magnitude improvement in cost per mile. Our target actually is that we have pet snail named Gary from....Sponge Bob Square Pants. Gary is capable of going 14 times faster than a tunnel boring machine. We want to beat Gary."

Is this not an attempt to just derail CAHSR, or HSR in general? Nearly every anti-HSR comment is now 'we should build hyperloop instead!'.

That's impressively cynical, but Musk obviously stands to gain a lot if people keep California using roads. California wants to go green and he sells the best technology for green road-based transportation.

To be fair, he also stands a lot to gain if Hyperloop becomes a reality.

How does that work? He had an idea and promoted it. No one's going to be paying him royalties for it.

He's also building one, so he has a vested interest in the technology being successful: https://techcrunch.com/2017/08/04/elon-musk-reportedly-plann...

heh.

Reminds me of when an actual AI expert pointed out that by making dire predictions about smart AI, that we are nowhere near having, keeps eyes off the autopilot system and the issues with that.

I'm not sure it really affects his electronic car business one way or the other. Firstly, he isn't necessarily selling Tesla's to people for commuting between SF and LA, that's well outside his range, and as fast as the supercharging is, it loses out to regular gasoline refilling. Secondly, I'm not sure the amount of people potentially served by the hyperloop is more than a blip in the sales charts for even a regular car company (as deciding factors for car purchasing decisions, it seems relatively low to me).

Okay, we'll bookmark this for the future. Amtrak estimated a HSR/"bullet train" from D.C. to Boston would cost ~$151 billion. Since LA to SF will only be two stops, and we can assume Amtrak estimated at least 4 (Philly and NYC) let's bring the number down closer to $100 billion to make it even.

Do we think the Hyperloop costs are more or less than that? 1/8 of a theoretical $100b is $12.5b -- is $12.5billion for Hyperloop unreasonable? I have no opinion on cost personally, and I don't know economics, but this $151 billion from D.C. to Boston was floated by the established US player in rail infrastructure. Anything less than that number is fantastic, right?

You cannot compare D.C to Boston with SF to LAX. The east coast has much denser population so that you need more tunnels. LAX to SF HSR would not cost $151bn. Start and end are expensive but the part in the middle is "relatively" cheap.

Right, for comparison's sake I chopped off $51 billion bucks from the Northeast Corridor projection in the interest of balancing out the two.

Just because I felt like doing some more Google-fu, I found an article in the LA Times[0] that says a California HSR/bullet-train project is going to overshoot it's original budget (and deadline) of $68 billion. There's a lot of info out there about this project, and it's potential overruns, but let's forget all that and just stick to the original planned cost: $68 billion, so, about half of the cost of Amtrak's northeastern bullet. Let's use that number for our comparison.

If we are going to hold Elon to his "we can do it for 1/8th the cost" blurb, then we are giving him like $8.5 billion to use for his SF/LA Hyperloop. Still, that doesn't sound unreasonable, right? He'll get a good deal on tunnel boring with his other company, and fuselage manufacturing can be handled by SpaceX. When it comes to financing, I don't think it's a major issue compared to other infrastructure works out there.

Like I said, though, I'm just going off the top of my head. I don't know finance or economics or vacuums or magnets.

There's a new LIRR train being dug in NYC called East Side Access. This is a commuter rail line, and is going to cost ~$10 billion or so[1]. Logistically these aren't the same, obviously, but if adding to the LIRR is worth ~$10b then surely the Hyperloop experiment is as well, right?

EDIT: Also, just for disclosure, I don't live in California and I'd probably never end up using the Hyperloop myself so I'm neither for nor against it versus any HSR. I just want to entertain the idea that cost shouldn't be the main focus of discussion IMO.

___

0. http://www.latimes.com/local/california/la-me-bullet-train-c... 1. http://web.mta.info/capital/esa_alt.html

> If we are going to hold Elon to his "we can do it for 1/8th the cost" blurb, then we are giving him like $8.5 billion to use for his SF/LA Hyperloop. Still, that doesn't sound unreasonable, right

I think boring company claimed it can bring down costs to 100 million dollars per mile (how? Just reduce tunnel size! SMH. Plus 100 million dollar a mile isn't too far from prices of current tunnels). Sf to la is 380 miles. That's 38 billion dollars just for the tunnel?

I think hyperloop is certainly feasible, if you spend hundreds of billions in it, the problems can be dealt with costly engineering.

If you think hsr will cost 150 billion, a maglev will be costlier, tunnels are extremely costly to build, but building a maglev in a vacuum tunnel will cost only 6 billion? How will cost savings of an order of magnitude happen?

> Okay, we'll bookmark this for the future. Amtrak estimated a HSR/"bullet train" from D.C. to Boston would cost ~$151 billion. Since LA to SF will only be two stops, and we can assume Amtrak estimated at least 4 (Philly and NYC) let's bring the number down closer to $100 billion to make it even.

You're starting with the wrong number. The $151B number is Amtrak's estimated cost for it's "shoot for the stars" plan. Its basic NEC stretch plan is effectively 4-track corridor (shared with commuter rail) from DC to Boston, with new stations and new inner city track being laid in Philly and Baltimore, two new Hudson River tubes and a new station annex in NYC, among other less notable improvements. I don't recall if the "abandon current Connecticut track and instead go from NYC along Long Island and tunnel under the sound" plan is in that tier or the next one. The stretch tier is somewhere in the region of $70B.

The reach for the stars tier is "do all that, and then build a parallel 2-track dedicated HSR track from DC to Boston." That's $150B. Which means building dedicated track from DC to Boston is only $80B. The estimated full buildout for CAHSR (including San Diego and Sacremento) is around $80B, as I recall.

Meanwhile Switzerland, one of the top most expensive countries in the world, has tunnelled at the base below the alps (one of the most difficult terrains to tunnel due to vertical sediment layers) for ~20B. Workers died on that project too, but that risk is simply being managed rather than grinding everything to a halt.

I'd argue it's absolutely possible to improve the cost on American public transit infrastructure by one to two orders of magnitude if you can simply get a way around government capture.

How much was there in land take for the Gotthard Base Tunnel? How much land take would there be for Amtrak's $80B line from DC to Boston?

My guess would be that the cost of land in the north-east, combined with AFAIK pretty strong ownership rights, makes it that expensive.

While American construction costs do seem to run higher than comparative costs in Europe and Asia, the multiplier is closer to something around 2-3×, not 10-100×.

The cost of the Gotthard Base Tunnel comes out to around $300-400 million / km. By comparison, the Second Avenue Subway comes out to around $800-1000 million / km (I'm subtracting a bit because the subway has three stops--which tend to be really expensive money pits--and the base tunnel does not).

Honestly, 12.5B look like a ridiculously small amount to me for an infrastructure project stretching along 550Km and involving an unproven, technically very challenging technology.

I'm not saying the hyperloop is a good or bad idea, but any suggestion that it would be cheaper than the equivalent HSR seems utterly insane. How could it possibly be cheaper? You have the same ROW requirements, exponentially more expensive track, and the overhead of the fact that it's brand new technology. The fact that the vehicle itself might be cheaper than an HSR train after the tech scales up is almost an afterthought next to those costs.

The track should be cheaper because hyperloop has (almost) no impact on existing infrastructures. Expropriation costs for train are huge.

How could the Hyperloop not require the same expropriations? A giant tube isn't any smaller than a train track. Is it the fact that it's elevated? Rail can be elevated. Yes, rail is heavy, but so are giant evacuated steel tubes.

Run it down the center of the freeway? Having stations for trains in the middle of freeways isn't great for pedestrians, but might be fine for long distance transport such as hyperloop.

HSR is supposed to be log distance transport. It's I meant to solve the exact same problems as Hyperloop, but by powerful engines and streamlined railcars instead of evacuated tubes.

Again, why not put HSR tracks in the same place you'd put a Hyperloop track? Put the station in the same place. The tubes aren't smaller or lighter.

I think he's planning on making lots of tunnels.

https://www.wired.com/story/elon-musk-test-tunnel/

Tunnels are very, very expensive.

e.g. https://pedestrianobservations.com/2016/12/15/second-avenue-...

> The total length of the project, from 96th Street and 2nd Avenue to 125th Street and Lexington, is about 2.7 km. At $2.2 billion per km, this sets a new world record for subway construction costs, breaking that of the first phase of the same line, which only cost $1.7 billion per km.

I read one article that blamed the high cost of modern subways primarily on the hyper-elaborate modern subway stations and the actual tunnels are a small component of the cost.

I think you're right - tunnels seem to be about $200-400M per km when considered purely on their own.

Musk seems to think he can do it faster/cheaper. His track record makes it worth considering.

The article I linked to smacks his arrogance down over costs that he's overlooking, but even that may be addressable. Let's see what he delivers in the next 5 years.

I don't think Musk want hyperloop to be built above the ground. This is just my own personal opinion, not sure if that's how it's going to work, but in one of his interview (TED? I think), he mentioned one of the reasons for his Boring company, was it would be relatively cheaper to build hyperloop - because the atmospheric pressure will be lower underground - and since you don't have to worry about building large infrastructure for only one purpose (hyperloop) - the cost will be much much cheaper. Because under ground tunnels can be build for both hyperloop and roads for cars and it scale better than above ground.

I don't think above ground is economically and technically feasible, there are just too many unanswered issue that almost no one has answer to. Not to mention it will be very very expensive and time consuming.

Why would atmospheric pressure be lower underground?

Tunnel boring is orders of magnitude more expensive than railway construction, at typically ~a few hundred million dollars per mile vs ~a few million dollars per mile. I know Musk has plans to magically make tunnel boring significantly cheaper but you need two orders of magnitude cost reduction to make the tunnel digging not cost more than an entire above ground railway.

Cost references: https://tunneltalk.com/TunnelTECH-Apr2015-Arup-large-diamete...

https://www.compassinternational.net/railroad-engineering-co...

Maybe you're looking in the wrong place for the punchline. Perhaps it will be delivered in the final bill for the HSR...

Everyone has been chuckling about electric cars and private rockets until recently, and that was by experts in the respective fields. Chuckle all you want.

Go here and scroll down to "Select historical production vehicles" - https://en.wikipedia.org/wiki/History_of_the_electric_vehicl...

I grew up in the 70s and saw production electric vehicles that are not on this list. People have been chuckling for a long time and for good reason. It's not that electric vehicles won't eventually be successful -- they must. It's that the time from, "Hey, an electric vehicle would be a good idea" to "Hey, we've figured out all the engineering challenges to make this viable" is longer than the lifetime of most of the companies that entered the fray.

Look at the list of vehicles from before 1990. That's the "getting to the market first" list. All gone. Then look at the list from 1990 to 2010. How many of those will be major players in the electric vehicle market? My point is that I'm relatively old for someone on HN and there have been production electric vehicles since well before I was born. They only started to be viable in the last few years -- and even then, we probably need a few more breakthroughs in battery technology before the market settles out.

Long distance, super fast trains in tunnels? https://en.wikipedia.org/wiki/Ch%C5%AB%C5%8D_Shinkansen I saw the tunnel the other day. I was surprised that practically the whole damn length of the maglev tracks will be in a tunnel, but it makes total sense. That's been under construction for 10 years. They are optimistically expecting to make a profit in 2026 when "costs stabilise".

Will long distance, high speed trains in low pressure tunnels be successful? I have no doubt. Will Hyperloop One be successful? Umm... frozen balls of ice remaining intact in a very hot place probably have a better chance, but I won't say they will definitely fail. If they can secure something like a trillion dollars and invest it wisely over the next 30-40 years, then their chances will improve significantly.

Let me be clear. I was talking about criticism of Musk specifically. Pretty much EVERYONE in the aerospace industry thought he was nuts, and yet SpaceX made it to orbit. Even a couple months before their first stage landing others were saying it was impossible.

An initial pipeline could be built with a lower target speed. This would postpone some of the potential challenges with regard to heat dissipation, and also route selection.

For example, you wouldn't have to worry so much about tunneling under a mountain to ameliorate the G-forces if you are traveling more slowly. Once you prove out the route (which could still be competitive with any existing train service), you incrementally build tunnels and modify the route to improve your target speed.

(As for heat discharge, Im pretty sure that was addressed in EM's original paper - dumping the excess heat into an on-board water tank which is exchanged at the end of the ride, or are you referring to something else?)

> Das ist nicht gute.

Should be "Das ist nicht gut".

People often, rightfully so, confuse the misappropriated name of Hyperloop One with Elon's Hyperloop idea, his of which I believe does have sound economics and certainly its pro list outweighs Hyperloop One's technology.

Could you point out the differences in both technological and economical approaches?

I'll try to remember to reply to this after the weekend, if not and you're still interested shoot me an email matt@engn.com

Reply on news.ycombinator.com