Cost of launches per kg will plunge

SpaceX's Starship fourth launch was a resounding success.  Both the booster and the ship came down for soft landings, even though one of the ship's fins was nearly burnt off by the heat of re-entry.  But compared with the first and second launches, when the Starship/booster combo exploded, and the third launch, when the Starship disintegrated during re-entry, these were huge successes.  Remember, SpaceX learns by iteration.  It tries something, then fixes any errors, then tries again, fixes new problems, tries again, and so on in a continuous program of improvement.  And the logic of the progress so far is that flight five (late July?) will be still more successful.  And flight six after that, and flight seven .....

By the end of this year, Starship will probably be carrying cargo to and from space.  And when it is, it will cut the cost of launching 1 kilo to orbit to around $20.  Musk says that each launch will cost $2 million, and each Starship can lift 100 tonnes into orbit.  Even if you allow for fat profit margins, making that, say $50/kg, it still means that a human weighing 100 kg could go to LEO (low earth orbit) for $5,000.  Though Musk says there will have to be hundreds of accident-free launches before Starship will be considered safe.  

That's just V-1 of Starship.  V-3, which will go into construction in a few months, will be even more efficient.   Its propellant load will increase 20%, but it will double its payload to 200 tonnes.  Cost per kg to orbit will fall to $12/kg.

The chart below, from Our World in Data, shows the cost of launching 1 kg to orbit since 1961, each observation adjusted for inflation since then.    SpaceX's first rocket, Falcon 1, cost $12,600/kg.  It was not re-usable.  The Falcon 9, which had a re-usable booster, but not second stage, cost $2,600/kg, half the cost of its nearest non-reusable competitor.  Falcon Heavy, three Falcon 9 boosters yoked together, cut the cost to $1500/kg.   And Starship V-3 will cut the cost to $20/kg.  Its data point will be right off the bottom of the chart.

It's impossible to know for sure how this will change the world.  But as Tony Seba says, a ten-times cost reduction leads to disruption and opportunity.   Since SpaceX started, it will have engineered a 250 times cost reduction for lifting one kilogram to LEO.   

This opens up the inner solar system to exploration.   The Moon and Mars will be in reach, affordably.  The cost of sending a single Starship to Mars will be $20 million in fuel (it will need 7 propellant ship launches to refuel it in space, each one carrying 200 tonnes of propellant).   Tripling that for food, life systems, etc., gives us a cost per ship of $60 million.  Even if we send 20 ships on the first expedition to Mars, with 10 astronauts per ship, with the rest of the payload devoted to food, shelters, water and air purification plants, rovers, and other things needed for survival on Mars, the total cost would be $1.2 billion.   That's less than half SpaceX's 2023 profit.  SpaceX could fund the first Mars mission with its own money, if it wanted to.

When asked to provide a costing for getting to Mars 20 years ago, in pre-SpaceX days, NASA estimated $100 billion (yes, with a b) for 5 astronauts, in then-money.   Things have come a long way since then.

Here's a forecast I believe

This is a chart from a chap called Wills, reported in RenewEconomy.  It shows demand for ICEVs (petrol/diesel/gasoline cars and lorries), for EVs and PHEVs (plug in hybrids).  Note that demand for ICEVs doesn't decline in a nice, safe linear trend.  It plunges.  It collapses.  It's a massive disruption. Now, it's possible some die-hards will still be buying ICEVs in 2026 and 2027, but they will be like those people who insist that vinyl is better than digital.  Except governments want to stop air pollution, so . . . .

Source

In the US, a car's average life is 11 or 12 years; globally it's higher (18 to 20) .  But EVs are so vastly superior to ICEVs that only the very poor will keep their old ICEVs as EVs flood the market.  Average life for aging ICEVs is likely to be 10 or 12 years or less, maybe much less, once EVs become the norm.  Which implies (see chart below) that total global oil demand (and CO2 emissions from oil) will start to fall by 5 or 5.5% per annum from 2026 onwards, and that rate of decline will accelerate as the global car and lorry fleet is modernised.  Even by 2024, it will be falling by 2 or 3% per annum.  This will be disastrous for oil companies and for those economies (Russia, Saudi, Kuwait, etc.) which have benefited so much from producing oil.  This has major military/political implications.

Prof Wills has other forecasts, about the peak in gas demand, for example, which I agree with.

Once again, it's truly excellent news for global CO2 emissions and global warming.  But this new technology will create losers, bigly.  Come to think of it, it will impoverish oil and coal billionaires and dry up their funding for reactionary right-wing parties.  Sad!

Norway nears 50% EV market share

4 years ago, plug-ins (PHEVs as well EVs) made up just 15% of Norway's new car market.  This percentage rose steadily.  In September, the ratio of plug-ins to all car sales reached 48%.

Now, there's been no change in the incentives Norway provides to buyers of plug-ins since 2014.  So what's changed?  Awareness, the spread of chargers, familiarity--and the reduction of incentives to buy EVs starting in January 2018.  Even if recent sales have been brought forward to take advantage of current generous incentives, it nevertheless still goes to show just how rapidly EVs and PHEVs will rise once the initial barriers of unfamiliarity and strangeness wear off.

Do not be misled by the low ratio of plug-ins to ICEVs (1.8%) in world car and light truck sales of today.  In 2012, plug-ins in Norway were just 3% of total car sales.  Now they make up nearly 50%.  Yes, the incentives in Norway to buy plug-ins are substantial.  But so was the cost differential between EVs/PHEVs and ICEVs when Norway started with its program to de-carbonise transport. And that cost differential is narrowing: within 5 or 6 years, EVs will cost pretty much the same as ICEVs.

There is every chance that world EV sales could rise very fast too.  Not as rapidly, because Norway is a single jurisdiction, and it's a small country in population terms, so new technologies and new ideas spread fast.  But California, already at 5%, could move very quickly (5 years) to 50%, even if the rest of the USA lags.  Sweden is already at 5%.  Other European countries are close to the flex point of the S curve, and plan future bans on ICEV sales.  And as always, China, 1/3rd of world auto output and sales, is determined to replace ICEVs with plug-ins to reduce pollution, and simultaneously grab for itself an even bigger chunk of the global car market.

Ironically, in Norway itself, the growth rate is likely to slow, in a typical "S"-curve way, as incentives to buy EVs are gradually reduced, starting next year.


[Read more here and here and here]

Source

A ticket with SpaceX for $2000

After his speech to the International Aeronautical Conference in Adelaide, South Australia, Elon Musk said about his earth-to-earth rocket flights, on Instagram:

"Cost per seat should be about the same as full fare economy in an aircraft. Forgot to mention that."

Extraordinary.  Melbourne to London in an hour for the same cost as a flight in an ordinary jet?  Once this technology is proven, who will want to fly on a conventional jet?  When I talked about this in my previous post, the guess I had in my head was a ticket cost of  $20,000, not something more like $1,000-$2,000.  The SpaceX ship could probably seat 500, on maybe 8 levels with the pilots in the nose (though in fact automated).  At this price, the cost of launch and landing (let's be conservative) will be $2,000*500 = $1 million.  (Is this just fuel and maintenance or did Musk include depreciation in his calcs?)

When it comes to Mars, though, the big cost of a trip won't be fuel and maintenance, it will be depreciation, because of re-usability restrictions.  Mars and Earth are only in alignment for a few months every two years, so each spaceship will only be used maybe 10 or so times one way, 20 times there and back over a 20 year life. Let's say the BFR costs $400 million (the cost of an Airbus A380).  That means each trip to Mars will cost $20 million in depreciation.  Cost of launch $1 mill.  Food and such for 100 people: 6 months * $1000 per month* 100 = $600K.  So total cost of each Mars trip $22 million plus contingencies and things I've missed $4 mill.  However some of the cost will be borne by the cargo.  So say $13 million for passengers--$130,000 per person.  Except, initially anyway, SpaceX will be giving a free return ticket.  But how many will actually make use of that ticket?  I wonder.  And how long will it continue to need to make that offer?

Can we doubt that Musk will get the cost of a journey to Mars below $100,000, given that this is still early days and that he is a genius at cutting costs?  The BFR will get bigger, the engines more efficient, the journey faster.  Once a regular biennial service begins, and competition gets started, costs will decline steadily.

Journeys to the moon, however,  will be far cheaper than journeys to Mars, because the moon is less than a day away (384,000 kms from Earth at the speed of the Earth-to-Earth rocket of 27,000 kph = 14 hours, give or take curving path, acceleration, deceleration, etc.), so the BFR could in principle be in almost constant use.  Unlike an Earth-to-Earth flight, the BFR would have to keep firing its rockets to accelerate from 27K kph to escape velocity (roughly 40K kph) and then decelerate to land on the Moon without the benefit of an atmosphere to slow it down.  Which means that to get to the Moon and back, the BFR would have to refuel in orbit round Earth (it couldn't, initially anyway, refuel on the Moon). Musk didn't mention what payload the BFR could lift off the Moon, but he did say that the BFR's payload from Mars would be 20-50 tons (depending on the different phases of the Mars and Earth orbits, I guess).  Gravity on the Moon is a bit less than half gravity on Mars, so, say 40-100 tons payload on the return journey from the Moon? So journeys to the Moon could cost perhaps 1.5 times the fuel with (say) 1/2 the total payload of a "flight" from Melbourne to London, or 3 times as much.  The cost of a business class ticket.  Did I screw up the calcs?  Let's say 5, nay, 10 times as much.  Still seriously, staggeringly cheap.  Seriously.  Stunning.  Remarkable.  World-changing. We've come a long way since Tintin and his friends journeyed to the Moon in 1954.

Hergé's vision of a moon trip in 1954.  (Source)

Musk's vision of a moon trip, 2017

(By the way, the ABC (Australian Broadcasting Corporation) has a nice piece on Musk's announcement to the IAC here, for those who don't like videos.)

[Update 22/08/19:  Using stainless steel for construction will cut the cost perhaps 10-fold, not just because it's 50 times cheaper than carbon-fibre composite, but because construction is so rapid, as evidenced by two Mark 1 Starships under construction and close to completion right now]

The meteorite

Tesla (and EVs in general) are going to disrupt many industries, probably faster than most people now think.

I liked this amusing cartoon from EVANNEX:

Only 1.2%?

At the beginning of major technological or market shifts, most people struggle to see that the nascent changes are going to cause massive disruption.  EV sales are only 1.2% of total car sales, they say.  How can 1.2% influence anything?

Take US EV (electric vehicle) and PHEV (Plug-In hybrid electric vehicles, i.e., those with a hybrid petrol/electric drive but with the capacity to also charge up their batteries from the grid).  Now as at December 2016, EV & PHEV sales made up 1.2% (seasonally adjusted: there is always a December spike in car sales, and weakness in January and February,  which you have to remove via seasonal adjustment) of total car and light truck sales.  The top chart shows EV sales (& PHEV, but I won’t keep on adding that; just remember that it includes both when I say ‘EV’ below) as a percentage of total car sales, and the second chart shows them in absolute terms.  (Source of basic data: Inside EVs)

Now notice a couple of things.

1. 6 years ago, EV sales were negligible.  (Incredibly, Tesla was still only selling the Roadster.)
2. From mid-2014 to mid-2015, sales fell—partly because the oil price collapsed, partly because the market was waiting for new EV models
3. During 2016, EV sales rose by 80%!  

In its Q4 2016 results presentation, Tesla announced that it was planning to produce 1,000 Model 3s a week in July 2017, 2,000 a week in August and 4,000 in September, rising to a peak of 10,000 in 2018.  Total EV sales in 2016 were about 160,000. If we add the likely Tesla Model 3 production in 2018, assuming no increase from any other manufacturer, total EV sales will increase to about 660,000 (a 4-fold increase) which will take EV sales to 5% of the total market.

But, wait, you say—what if Tesla doesn’t manage to sell 10,000 Model 3s a month?  Well, they have nearly 400,000 paid reservations.  OK, but what about Tesla’s production for overseas markets?  I don't know what percentage of Model 3 production Tesla has reserved for foreign markets.  But I suspect that most of the initial deposits were from US customers, and they will get priority.  And, remember, I’m assuming that sales by other manufacturers won’t rise.  But in fact there’ll be a new Nissan Leaf, a new VW e-golf and of course GM’s Bolt, all at around $30-$35K sticker price. It seems that every man and his dog will be offering one (or more) EV/PHEVs.  So I'm pretty comfortable with forecasting EV sales from all manufacturers will equal or exceed 5% of total car sales in the US in 2018.  By 2020, battery costs will be below $100/kW, and the cost of EVs will have fallen another 10 or 15%.  They will have the same or better sticker price as ICEs (cars with an internal combustion engine). So sales as a percent of total car sales will go on rising after 2018.

The key question is: what will this do to petrol (gasoline) sales?  Assuming EV sales rise as a percentage of total car sales by 5% a year, this is roughly what the collapse in demand for petrol sales will be, shown in the table below.  For convenience, I'm treating all sales as if they were of EVs with no PHEVs, but in effect, the expansion in battery sizes will move PHEVs to being EVs for every day and only hybrids on long trips.  I'm assuming a linear not an exponential transition.  Also, this is just for cars, but presumably sales of light and heavy trucks and busses will be affected similarly.  And, it's just for the US, but again, the global trends will be similar.  Electric bus sales are already 20% of total Chinese bus sales, for example.  I'm assuming that the average age of cars and light trucks remains unchanged at 11 years.

In the first year, petrol sales fall just 0.5%.  In 2019, another 1%.  By 2025, sales will be down a cumulative 15%.  By 2030 they will have halved.  But the oil crash will come before that.  As Bloomberg says (in this interesting piece):

We found that electric vehicles could displace oil demand of 2 million barrels a day as early as 2023. That would create a glut of oil equivalent to what triggered the 2014 oil crisis.

That's just 5 years away.  Actually, it's potentially worse than that.  Saudi Arabia has the world's largest reserves and the lowest cost of production.  Once it becomes clear to them (and the others in a similar position, such as Iraq and Iran) that it's a case of sell your oil now, or never, they will expand production to its maximum.  Why restrain production to keep the market "orderly" when high price competitors, (like fracking in the US, for example) will gain, and you one day won't be able to sell your oil at all?

I don't know when the oil crash will start, but once it does it will avalanche.  Each year the annual decline in demand will accelerate.  But wait : will falling petrol prices stop EV sales?  Maybe, but I suspect not, because EVs are just so cheap to run and so much more fun too, and they will just keep on getting cheaper.  Anyway, to reduce air pollution and CO2 emissions, some European countries and India plan to ban new ICE sales from 2025 on.

By 2025 or soon after, petrol service stations will start to close, unless they transition to EV charge stations.    By 2030, it will probly become hard to find an old-fashioned service station that still sells fuel.  That's just 13 years away.

In the year that Elon Musk's SpaceX will start the colonisation of Mars, Musk's other business will have shifted the world half way towards zero carbon emissions.  Interesting times.  I hope I live to see it all happen.

Clean disruption

This remarkable video by Tony Seba shows how dramatically the energy and transportation landscapes are going to shift over the next 5 to 10 years.  I don't usually post videos, because they take a lot of time to watch, more time than the equivalent text.  But I make an exception with this one.  Almost every minute of it is worth watching, and its conclusions are very significant for our society.  Watch the whole thing, then if you want to go back and see individual sections, refer to the notes below the video.

Some key projections and points, to whet your appetite, with my comments in square brackets:

• The decline in battery storage costs is accelerating [this was before Tesla announced that costs of the Powerwall had halved in one year]
• By 2020, one day's storage would cost $1 per day, less than a cup of coffee.
• Large scale storage will completely displace gas peaking power plants, soon.  Already in some locales they are cost effective even at $350/kW storage, but by 2020 storage will be $200/kW and by 2024 $100/kW.
• 32% of generating assets are used for just 6% of the time.  Batteries will be much cheaper [but we might need more because of weather related variability]
• The plunge in battery prices means that by 2017 an "average" EV will cost $35,000, by 2020 $30,000, by 2022 $22,000.
• EVs have only 18 moving parts compared to an ICE (internal combustion engine) vehicle with 2000+.  They are 5 times more efficient than an ICE and 10 times cheaper to run.
• By 2025 100% of all new vehicle sales will be electric
• By 2020 all EVs will have self-driving technology.  Already self-driving is feasible for 90% of the time.
• At some point most people won't own a car but will share in a self-driving, automated EV fleet [like those robo-taxis SF authors used to talk about] 
• Installed base of solar has doubled every 2 years since at least 1990.  We are just 7 doublings away from solar providing not just 100% of all electricity but of all energy.
• The cost of rooftop solar will soon fall below the cost of transmission which means no alternative power source will be economically efficient, not nuclear, not fusion, not hydro, not coal nor gas.
• But we will still need utility-scale power for factories, data centres, smelting metals [and blocks of flats and offices and EVs] 
• Solar at 5.5 cents/kWh is the equivalent of oil at $10 barrel and gas at $5/MMBtu [recent solar contracts are now 40% to 50% below 5.5 cents/kWh]

Notes:

At 2;35, how ATT which invented many of the key mobile phone technologies asked McKinsey and Co how many cell phones would there be in the US in 15 years (2000).  McKinsey's answer: 900,000.  Actual number 109 million.

At 4:10, how the internet was never going to be anything important.  "The internet will catastrophically collapse in 1996" (Robert Melcalfe 1995) "There is no reason anyone would want a computer in their home" (Ken Olsen, 1977)  "I do not believe the introduction of motor cars will ever affect the riding of horses" (Scott-Montague, 1903)  Seba points out that it is the experts or insiders who will deny that disruptive opportunities and risks.

At 4:44, why do smart people at smart organisations consistently fail to anticipate or lead market disruptions?

Exponential technologies.  At 6:56, Moore's Law: for the same dollar we get twice the  computing power every 2 years.  If you double every 2 years, over a decade that's a 1000 fold improvement, over 2 decades it's a 1 million fold improvement and over 3 decades a 1 billion fold improvement.  Kryder's Law: hard disk $ cost down 50% every 18 months.  Hendy's Law: digital imaging (pixels per $) down 59% every year.  Butter's Law of photonics (network capacity): the $ cost of transmitting one bit falls by 50% every 9 months,  The convergence of all 4 led to the smartphone and internet revolutions.

Key exponential technologies, at 8:56.  Sensors (9:52) market up 1000 times, costs down 1000 times, power down 1000 times, physical size down 1000 times in just 7 years.

Energy storage, at 11:38.  From 1995 to 2010, lithium-ion batteries fell by 14% per year.  Then 2 new industries came into lithium-ion: transportation and energy storage.  This accelerated the decline in the cost curve, from 2010 to 2014, cost declines accelerated to 16% per annum.  (Note that Seba's presentation was made in March 2016: since 2014 battery costs have halved)  At 16% by 2020 storage will cost $200/kW, by 2024 $100/kW.  Tesla's Powerwall 1 and Powerpack 1 battery costs were already (at the time of the presentation) below Seba's projected cost curve.  Now of course they are much lower than his curve.  Tesla got a billion dollars in orders on the announcement, and as a result the gigafactory, already planned to double world output of lithium-ion batteries, will be expanded in size by 48%.  at 15:58, Tesla isn't the only one doing this: BYD, Foxconn, LG Chem, Samsung SDI, TDK, Apple, Bosch, VW, etc.

Business model innovation, at 16:46.  Storage as a service 17:25.  By 2020, one day's storage (which by the way is more than we need) will cost about $1 (18:38) .  Half the cost of a coffee.  At 20:23, large scale storage likely to replace peaking power plants : billions of dollar in power plants used for just a few hours a year.  According to ConEd, 32% of generating assets used for just 6% of the time,

EVs, 22:10.  Tesla Model S: the best car ever made, not just the best EV.  The best selling large luxury car in America.   At 23:26, The EV is 5 times more efficient than the internal combustion engine (ICE), is 10 times cheaper to charge (24:08), has 2000+ moving parts compared to 18 in an EV (24:47), which makes EVs 10 to 100 times cheaper to maintain, and has much more torque (25:35).  When will EVs replace ICEs (26:56)?  $35,000 EV by 2017,  $30,000 by 2020, $22K by 2022.  By 2025 all new vehicles will be electric.  And note: it won't just be the incumbent companies who do it--Foxconn is to make an EV costing $15,000 (30:58)

Autonomous vehicles (32:52).  Tesla capable of self-driving 90% of the time (33:08).  What an autonomous car sees (34:02) Cost of LIDAR sensors has fallen  from $70,000 in 2012 to $10K in 2013 to $1K in 2014 to $250 in 2016 and a projected cost of $90 in a year or so (34:42)  World's first 1 teraflops computer in 2000 cost $46 million (36:13).  Would now cost $59 (36:43)  Is the market ready?  Yes, in Brazil, China, India, where traffic congestion is horrendous, self-driving cars are very desirable (37:25)  In San Francisco, 50% of Uber rides are car pools (38:43)  Cars are parked 96% of the time (39:09).  The convergence of EVs, self-driving, and shared cars will mean the end of individual car ownership (41:17)

The solar disruption (43:28) Installed base of solar has doubled every 2 years since at least 1990 (44:00)  7 more doublings, or 14 more years until solar can provide 100% of  total energy, not just electricity (44:25)  Grid Parity or God Parity (46:03)?  80% of global market will reach grid parity by 2017 (46:15)  Colour TV technology adoption curve (46:47) God Parity (48:18) where cost of rooftop solar is cheaper than costs of transmission, implying that no matter what alternative technology is used, even if it has zero cost, rooftop solar would be cheaper.  And that will happen by 2020 (49:16)  But we will still need utility scale solar, because blocks of flats, office blocks, aluminium/steel smelting (50:29)  Solar at 5.5 cents/kWh is the equivalent of oil at $10 barrel and gas at $5/MMBtu (51:00)