An important article...

louis · 2020-03-06 19:35:50

A great article in Ars Technica about Musk, Starship and Mars colonisation:

https://arstechnica.com/science/2020/03 … ttle-mars/

Full of insights.

The Starships are being manufactured from 2 metre high 301 steel barrels. Space X can now make "...two barrels a day, and it aims to reach a production cadence of four barrels a day."

2 barrels is 4 metres in height. So in a week they can make 28 metres in height. They are aiming to get to 56 metres in height in one week.

I am feeling good about all the news coming out of Boca Chica. I particularly like the sense we are now getting that Musk knows what he is doing - creating a covered production line for Starships at the site. That makes a lot of sense.

Here's a great quote from the article: "And now, in South Texas, Musk is getting close enough to Mars that he can almost taste its red dirt."

Here's another great quote, from Musk himself:

"The conventional space paradigms do not apply to what we’re doing here. We’re trying to build a massive fleet to make Mars habitable, to make life multi-planetary. I think we need, probably, on the order of 1,000 ships, and each of those ships would have more payload than the Saturn V—and be reusable.”

They are aiming to produce each Starship for $5 million. That is insanely low...Does anyone have an idea on how that might translate in cost per Kg to Mars?

If a Starship can operate for 100 flights (it's supposed to be fully reusable), you are spreading that construction cost over 100 flights = $50,000 per flight! Maintenance costs I would suggest will be high per flight. Maybe $100,000?? Then you have to crew the flight with professionals...That will vary as to whether it's human or non-human cargo. Let's go with cargo. Maybe that will be 6 crew at about $600,000. Plus you have to fuel the flight...I might have misremebered this, but I think that's about $500,000 to fuel a Starship.

Total costs might be only about $1.25 million per flight!

Of course, then you have to factor in the fuel to LEO flights - might be 5 of those. So maybe about $0.75 million per flight = $3.75 m

So total cost would be $5 m per flight. That would be

With a 100 ton payload, that would be $50 per kg!!! That would be a huge, huge game changer. The air cargo cost from China to USA is $3 per kg...

Yes it would be much more expensive to ship things from Mars but once you start factoring in no taxes, no land costs etc the $50 figure comes down significantly.

kbd512 · 2020-03-07 01:30:26

Louis,

The only real "game changer" would be accepting how physics and economics work in the real world. As much as I admire the beautiful visions of our daydreamers, those old physics and economics problems sure do seem to rear their ugly heads quite often. Unfortunately, I think basic economics is sufficient to deconstruct this latest "vision of the future".

I do have a better question, though. What makes Falcon Heavy 45 TIMES as expensive to construct and operate with 40% of the payload performance, 2/3rds as many smaller and simpler engines, equivalent fuel costs per kg of payload delivered, and Aluminum alloys that are only 3 times as expensive to fabricate parts from as compared to steel (and only 70.6t of structural mass vs 420t or 500t of dry mass, depending on which numbers you choose to believe), as compared to Starship Super Heavy?

There must be some crazy magical math going on there. It can't possibly be the fabrication costs, nor fuel costs, nor cost savings from re-flight, as we're about to show.

To go anywhere else, Starship requires another 5 tanker flights in quick succession. You're claiming 6 space launches for less than $4 million, but in practice SpaceX has never managed to come anywhere close to that number for a single mission that didn't require building and launching and maintaining another 5 rockets and a separate upper stage variant. Their Falcon 9 series of rockets cost $4,000/kg just to go to LEO once with a reusable booster. Falcon Heavy can launch 63.8t to LEO with no reusability, for a cost of about $2,350/kg, or 40t $2,250/kg with full booster reusability. In simple terms, hardly any money was saved by the purchaser of launch services from SpaceX implementing reusability, at least on a per-kilogram-of-payload basis. Maybe it doesn't cost them that much to refurbish the rocket and most of that is pure profit, in which case I'm really in the wrong business. However, Elon Musk himself said it's only a 30% cost savings over fabricating a new booster.

For some reason, you're here claiming that we might get one of those kilograms of payload to Mars, presumably from reusability, for an absolute minimum of 45 TIMES less cost.

Fun fact about steel vs Aluminum fabrication costs. Aluminum was only about 5 times as expensive to fabricate parts out of, as compared to steel, two decades ago. The costs to fabricate components from Aluminum has dropped substantially since that time. I think it's presently around 3 times as expensive, as compared to steel fabrication. The fuel cost for the Falcon series of rockets is slightly above $1.20 per kg and LNG is only slightly more expensive at around $1.30 or so per kg. Anyway, the fuel costs per kg are so close that there's no meaningful difference. Let's just round and call that $1M just for the 813t of LNG. LOX is pretty darned cheap, but that's still around $250K or so. Anyway, the fuel bill alone is at least $7.5M, ignoring any test firing requirements.

China sells 2mm thick 301 sheet is about $2,000 per ton. I seriously doubt anything much thinner than that would be suitable and that stuff flops around like rubber so there must be some serious internal stiffening applied. That's about $600K for all of the structural steel required to fabricate a complete vehicle. Shipping the product from China to the US is likely to run $1,600 per full container load to the port of Los Angeles. The legal weight limit is around 20t per container, so 15 full container loads at around $24,000 to LA. That's an average to LA and includes import duties and taxes, but it'd probably be pretty difficult to do substantially better than that. That takes about 3 weeks, on average. Air freight costs from China would be absurd (around $1,650,000 for 300t), but I've never heard of anyone air freighting sheet steel, either, but it normally takes about a week (the entire process, not the flight).

The cost to ship 300t about 1,200 miles by rail (AMTRAK's route from Los Angeles, CA to San Antonio, TX) is around $18,000 and would take around 35 hours. The cost to ship 300t from LA to San Antonio by truck would be about $54,000, but would arrive in around 19 hours. I seriously doubt that the extra 16 hours or so makes much of a difference to construction so long as you keep the steel rolling in. Last mile shipping would almost certainly be by truck. Those rates vary quite a bit, especially for hot shots (it frequently gets a little crazy, often comparable to air freight rates). That said, I'm quite certain that they'd just choose the cheapest method and keep the product rolling in.

The labor costs for welding stainless is, well, not that cheap. That said, I feel for whomever has to weld all this stuff together. In any event, using a series of roll forming dies they can probably fabricate almost every structural part on both vehicles except for the engine mounts and landing gear from a single gauge of material by doubling up the material or welding in support structure, which we frequently do in amateur built aircraft construction to limit the quantities of unique gauges of metals required and therefore cost. The engine mounts are likely 4130 Chrome-Moly tubing. We use that stuff in aircraft engine mounts for its fatigue resistance and strength after its been hardened, which is better than any reasonably priced stainless if cost is a consideration and lighter in weight for equivalent strength if weight is a consideration. 4130 is not terribly expensive, but certified material is not nearly as cheap as sheet steel, either. The labor cost associated with aircraft grade welding jobs is what makes the mounts so expensive. The engine mount material itself is generally less than 5% of the total cost. I'd love to know who he has out there welding who will fabricate the barrel sections for such low pay that they could ever make Starship only cost $5M, just in terms of labor.

Now for the bad news. SpaceX is claiming they're going to make 42 engines in the same thrust class as the RS-25 (Space Shuttle Main Engine) for around $95,000 a pop. Someone is claiming they're going to construct 400,000lbf thrust rocket engines for less than half the price of a 400hp Lycoming IO-720, which run north of $230,000 brand new and $195,000 rebuilt. I'm calling BS on that one.

I'd be floored if the cost of a Raptor engine was less than several million dollars a pop even if the labor was essentially free, which it's not. A Pratt & Whitney PT-6A is $500,000 to $1,000,000 per copy, but as many parts as those have, they're just toys compared to full flow staged combustion rocket engines. PT-6A's are typically 750hp to 1,200hp gas turbines and the factory has made well over 50,000 thousand of them. They weigh around 300 pounds. The turbo pumps on Raptor probably weigh more than that and are many times more powerful. This new Raptor rocket engine is probably 10 times as heavy and made from similar metals and fabrication techniques, assuming it's built to last.

SpaceX could take over the aircraft engine industry in its entirety if he can actually make rocket engines that cheaply.

tahanson43206 · 2020-03-07 06:06:42

For Louis re your reply to kbd512 #2

Please do not quote the entire long post by kbd512 in your reply.

Simply refer to the post number.

When you do reply, please take the points kbd512 has made and either confirm or refute them.

Please do not fall into the trap of attempting a hand wave to dismiss the points kbd512 has made.

(th)

louis · 2020-03-07 07:13:16

kbd512 wrote:

Louis,
The only real "game changer" would be accepting how physics and economics work in the real world. As much as I admire the beautiful visions of our daydreamers, those old physics and economics problems sure do seem to rear their ugly heads quite often. Unfortunately, I think basic economics is sufficient to deconstruct this latest "vision of the future".

Physics is one thing - there are only a limited number of ways of getting a craft to defy gravity and you can take it as agreed that currently that means we rely on combustion of rocket fuel. Economics is an entirely different matter.

I do have a better question, though. What makes Falcon Heavy 45 TIMES as expensive to construct and operate with 40% of the payload performance, 2/3rds as many smaller and simpler engines, equivalent fuel costs per kg of payload delivered, and Aluminum alloys that are only 3 times as expensive to fabricate parts from as compared to steel (and only 70.6t of structural mass vs 420t or 500t of dry mass, depending on which numbers you choose to believe), as compared to Starship Super Heavy? There must be some crazy magical math going on there. It can't possibly be the fabrication costs, nor fuel costs, nor cost savings from re-flight, as we're about to show.

As far as I know they have only ever built three or four Falcon 9 Heavy craft and don't have plans to expand production hugely. The Starship will supercede it in due course. I suspect that's what makes it so much more expensive. The Falcon Heavy 9's development was a tortuous thing, and presumably added hugely to the costs of the project. I am presuming you are referring to the total cost to date and not the cost just to build one. In terms of what is charged per ton, I guess they go with what the market, such as it is, will take.

I always said that Falcon 9 Heavy was a wrong turning, a detour that took them to a dead end. Starship is an entirely different animal, not least because it is multi-use.

To go anywhere else, Starship requires another 5 tanker flights in quick succession. You're claiming 6 space launches for less than $4 million, but in practice SpaceX has never managed to come anywhere close to that number for a single mission that didn't require building and launching and maintaining another 5 rockets and a separate upper stage variant. Their Falcon 9 series of rockets cost $4,000/kg just to go to LEO once with a reusable booster. Falcon Heavy can launch 63.8t to LEO with no reusability, for a cost of about $2,350/kg, or 40t $2,250/kg with full booster reusability. In simple terms, hardly any money was saved by the purchaser of launch services from SpaceX implementing reusability, at least on a per-kilogram-of-payload basis. Maybe it doesn't cost them that much to refurbish the rocket and most of that is pure profit, in which case I'm really in the wrong business. However, Elon Musk himself said it's only a 30% cost savings over fabricating a new booster.
For some reason, you're here claiming that we might get one of those kilograms of payload to Mars, presumably from reusability, for an absolute minimum of 45 TIMES less cost.

The Starship won't always need refuelling. It won't need refuelling for satellite launches, orbital tourism, ISS resupply or Earth-to-Earth transport.

I did neglect the ground control and spaceport maintenance costs which will be substantial, but again if you can get up to 3 flights per day, as Musk wants, those costs will be spread over many more flights than is now the case. So woudl probably not add a huge amount. For a ball park figure I'd say 200 GC staff (working shifts) at $200,000 per person (salaries plus on costs) would give you $40 million. Let's add $60 million for spaceport costs to bring it to $100 million .With 3 flights per day that would be $31000 per flight. So $186,000 for 6 flights.

So the main reasons for Starship being much cheaper are: 1. It's going to be a much simpler beast - a two stage rocket not a 3 rockets strapped together affair. 2. It's going to be fully reusable. 3. Because so many Starships are going to be produced, economies of scale kick in and the cost per unit falls hugely. 4. Because there will be so many flights the costs per flight are going to be much smaller.

Fun fact about steel vs Aluminum fabrication costs. Aluminum was only about 5 times as expensive to fabricate parts out of, as compared to steel, two decades ago. The costs to fabricate components from Aluminum has dropped substantially since that time. I think it's presently around 3 times as expensive, as compared to steel fabrication. The fuel cost for the Falcon series of rockets is slightly above $1.20 per kg and LNG is only slightly more expensive at around $1.30 or so per kg. Anyway, the fuel costs per kg are so close that there's no meaningful difference. Let's just round and call that $1M just for the 813t of LNG. LOX is pretty darned cheap, but that's still around $250K or so. Anyway, the fuel bill alone is at least $7.5M, ignoring any test firing requirements.

Happy to take your advice on fuel costs, I obviously misremembered.

China sells 2mm thick 301 sheet is about $2,000 per ton. I seriously doubt anything much thinner than that would be suitable and that stuff flops around like rubber so there must be some serious internal stiffening applied. That's about $600K for all of the structural steel required to fabricate a complete vehicle. Shipping the product from China to the US is likely to run $1,600 per full container load to the port of Los Angeles. The legal weight limit is around 20t per container, so 15 full container loads at around $24,000 to LA. That's an average to LA and includes import duties and taxes, but it'd probably be pretty difficult to do substantially better than that. That takes about 3 weeks, on average. Air freight costs from China would be absurd (around $1,650,000 for 300t), but I've never heard of anyone air freighting sheet steel, either, but it normally takes about a week (the entire process, not the flight).
The cost to ship 300t about 1,200 miles by rail (AMTRAK's route from Los Angeles, CA to San Antonio, TX) is around $18,000 and would take around 35 hours. The cost to ship 300t from LA to San Antonio by truck would be about $54,000, but would arrive in around 19 hours. I seriously doubt that the extra 16 hours or so makes much of a difference to construction so long as you keep the steel rolling in. Last mile shipping would almost certainly be by truck. Those rates vary quite a bit, especially for hot shots (it frequently gets a little crazy, often comparable to air freight rates). That said, I'm quite certain that they'd just choose the cheapest method and keep the product rolling in.
The labor costs for welding stainless is, well, not that cheap. That said, I feel for whomever has to weld all this stuff together. In any event, using a series of roll forming dies they can probably fabricate almost every structural part on both vehicles except for the engine mounts and landing gear from a single gauge of material by doubling up the material or welding in support structure, which we frequently do in amateur built aircraft construction to limit the quantities of unique gauges of metals required and therefore cost. The engine mounts are likely 4130 Chrome-Moly tubing. We use that stuff in aircraft engine mounts for its fatigue resistance and strength after its been hardened, which is better than any reasonably priced stainless if cost is a consideration and lighter in weight for equivalent strength if weight is a consideration. 4130 is not terribly expensive, but certified material is not nearly as cheap as sheet steel, either. The labor cost associated with aircraft grade welding jobs is what makes the mounts so expensive. The engine mount material itself is generally less than 5% of the total cost. I'd love to know who he has out there welding who will fabricate the barrel sections for such low pay that they could ever make Starship only cost $5M, just in terms of labor.
Now for the bad news. SpaceX is claiming they're going to make 42 engines in the same thrust class as the RS-25 (Space Shuttle Main Engine) for around $95,000 a pop. Someone is claiming they're going to construct 400,000lbf thrust rocket engines for less than half the price of a 400hp Lycoming IO-720, which run north of $230,000 brand new and $195,000 rebuilt. I'm calling BS on that one.
I'd be floored if the cost of a Raptor engine was less than several million dollars a pop even if the labor was essentially free, which it's not. A Pratt & Whitney PT-6A is $500,000 to $1,000,000 per copy, but as many parts as those have, they're just toys compared to full flow staged combustion rocket engines. PT-6A's are typically 750hp to 1,200hp gas turbines and the factory has made well over 50,000 thousand of them. They weigh around 300 pounds. The turbo pumps on Raptor probably weigh more than that and are many times more powerful. This new Raptor rocket engine is probably 10 times as heavy and made from similar metals and fabrication techniques, assuming it's built to last.
SpaceX could take over the aircraft engine industry in its entirety if he can actually make rocket engines that cheaply.

I can't comment on these other claims of yours. You are second-guessing Musk, the owner of Space X which has already brought down rocket costs substantially. If you are basing your costs on one off purchases, that is obviously going to be misleading and of course they will include a series of profit margins (whereas Space X is making the items themselves). You have to ask how much you can get prices down if you have a production line of 3 rockets a week - something which has never been attempted before.

Taking on Musk's estimate and amending for ground control, spaceport and additional fuel costs as discussed above, that brings the revised total for a Mars launch to $9.2 m or $92 per kg. It will probably be higher because you have to factor in Spaceport costs on Mars and perhaps an extended role for a dedicated mission control team on Earth and on Mars (but Earth-Mars transit should become fairly standard). Let's call it a round $100 per kg. Still a game changer.

Last edited by louis (2020-03-07 07:16:07)

GW Johnson · 2020-03-07 09:47:21

Well, look at this another way: are we doing better now than we were before?

Apollo era: $30B in 1970's $ to land 12 men on the moon. Allowing for inflation that's close to $100B in today's $. There were 7 Mercury flights, about a dozen Gemini flights, and about a dozen Saturn flights counting the tests. That's around $100B/2020$ for some 30 flights, or about $3B per launch.

The verified costs of the Space Shuttle were also right at $1B per launch in 1990 $. In today's $, that's probably no more than $2B/launch, because inflation hasn't been so severe in recent decades.

NASA with SLS/Orion is doing what it did during Apollo with Shuttle tinkertoys. They have yet to fly after two decades development, but the projected price for doing it the old way is $1-2B/launch in today's $. Why should that be surprising at all? The initial version of SLS is said to be capable of delivering 70 tons to LEO. That's roughly $21M/ton to LEO, give or take a big uncertainty.

ULA no longer flies the Delta-4 Heavy. They never made any attempt to reduce launch costs with it. As a result, it only launched military items in the last years it was used, at something like $15-20M/ton, about like NASA.

ULA did attempt to reduce costs with its Atlas-5, in order to compete with Spacex. Depending on which version, how many engines in the Centaur stage, and how many SRB's, their price per launch is in the $100-130M/launch range in today's $. Order of magnitude lower than NASA's track record, but a lower payload, too. About 20 metric tons to LEO. Call it $5-6M/ton to LEO. Yep, they're doing it a lot better than NASA ever did, by a factor of 3 or 4!

Spacex has been driving the competition in the commercial launch market, which is the motivation for lower launch costs. Falcon-9 now delivers almost 20 tons max to LEO (closer to 10 if you recover the booster) for a published price of $63M/launch not-recoverable. I haven't seen anything on the price break for recoverable-booster flights; take a guess and call it $50M. You're looking at something in the $3-5M/ton range for LEO. That's even better than ULA, and way to hell-and-gone better than NASA.

Spacex's Falcon-Heavy has only flown twice, but is listed on the website at around $85M per launch, and 63 tons to LEO, flown nonrecoverable. That's around $1.3-1.4M/ton to LEO. Remarkable! Yep, size matters: the bigger ship carrying more cargo really is cheaper to build and operate, even if it's a spaceship, not an ocean-going ship.

This stuff is not a linear curve, it's a curve of diminishing returns. I would expect the Starship/Super Heavy to have an even lower price per ton to LEO than Falcon Heavy, just because it is so much bigger than Falcon Heavy. Not dramatically lower (orders of magnitude, that's nonsense), but maybe around $0.5-1M/ton to LEO, once they get past all the teething troubles they are going to have with it.

I ran some numbers on this Starship/Super Heavy system, based on the various presentations that have been made public by Musk. I think it takes 6 not 5 tanker flights to fully refill a Starship on-orbit in LEO. And you don't go anywhere useful without a full refill. Not the moon, not Mars, not anywhere else.

Now I am likely wrong, but on the overoptimistic side. Spacex has yet to come anywhere near the vehicle inert mass they were projecting when I made those estimates. It might be a lot more tankers than 6 to refill, and I'd bet real money the payload to Mars will struggle to ever be 100 tons.

They think they're having teething troubles with tank welds now -- wait till they actually fly and start crashing or exploding vehicles. This spaceflight thing is not for the faint-of-heart or the low-of-budget, even when doing it orders and orders of magnitude cheaper than NASA.

Sorry to rain on anybody's parade, but those are my best projected guesses, based on 2020 $.

GW

Last edited by GW Johnson (2020-03-07 10:06:15)

SpaceNut · 2020-03-07 09:52:12

Replies for Louis, reading comments first then article...

Since sheets of stainless would be easier to use for shaping a ship it sounds to me like the use of Barrels is a recycling of materials which are junk value. No wonder its costs are under the Falcon 9 or of its heavy....Now if you are using the "barrel" as a term for the cylinder shaped sections of the outer shell then they must be using a lot of cheap stainless not from the US.

reading from the photo, which is the fuel tank construction image where the term "barrel" is used is speaking of sections of the construction assembly. The dome is made from shaped cut pieces that are welded together which accounts for failure along the seams. Little wonder why as the number goes up so will the rate of failure. Normal tanks are a single unit constructed tested and then placed into an assembly where its going to be used.
Trying to cheat material mass with this construction technique of intergrated piece assembly increase risk for failure of the tank as it can not be proven to be good as the shell is part of its assembly rather than a seperate part of the structure of the ship. Its also the reason for needing more man hours to build in a shorter period of time for that large piece cutting, shaping and welding build.

A starship can not fly 100 flights as the trips to mars ( 2yr approximate) are not an up and down trip with not to meantion heatshielding wear will mean replacing if you do keep the majority of the vehicle.

The first stage BFR will not do 100 flights as well since we have predictable falcon cores with a reuse rate of just 10 not counting landing errors. So no 100 flights with these either.

GW Johnson · 2020-03-07 10:18:57

One of the problems Spacex is having building its Starship prototypes is exactly what Spacenut mentioned: having to weld a bunch of curved panels together to make a cylindrical section ("barrel"). That's because this thing is about 10 m in diameter, and there's no stock available big enough to make those sections in one piece, nor the machines big enough to shape and fabricate it.

What that means is you will get a lumpy, mis-shapen-looking product, and it will have a lot of welds whose reliability you must question. The lumpiness is not that big a problem in large objects, because it is a small percentage of vehicle dimension. It just looks ugly to the eye. The weld reliability is proving to be a problem for Spacex, though. I haven't seen anything yet to indicated that they understand the nuances of how to design around that.

Teething troubles. They will get past it, but it will cost them $ and it will cause delay. That second is the difference between time and Musk time as it relates to schedules. About a factor of 3 on the originally-projected interval.

Doesn't matter if Musk doesn't reach Mars until the mid-to-late 2030's. He will still beat NASA there. They will still be circling the moon instead of going to Mars. If they actually circle the moon at all.

GW

kbd512 · 2020-03-07 16:36:43

Louis,

I have worked in supply chain for consumer products for a number of different major manufacturers, all of them global operations, since 2007. They were involved in the manufacture of everything from foodstuffs, beverages, pharmaceuticals, sports apparel and equipment, computers and other consumer electronics, various kinds of steel or steel parts, and now for an oilfield services company that makes its own tools and purchases mass quantities of materials to drill for oil and natural gas, both here in America and around the world. That's why I know a little bit about what it costs to get something from Point A to Point B, at least here on Earth. Everyone I've worked for is moving ships and rail cars and trucks filled with raw materials and finished good products.

In my personal life, I have looked high and low for that true bargain of the aircraft engine world. All I can tell you about that is that there aren't many bargains to be had. Anything manufactured in the low thousands using highly specialized machining methods, certified materials, multiple material and finished part inspections, and lots of touch labor is inherently expensive. All aircraft engines, which includes rocket engines, will remain very high dollar items as a function of low production volumes and high labor costs unless something breaks that equation.

Fabricating 1,000 vehicles with 42,000 engines does zip / zero / zilch / nada to break that equation. If Elon Musk truly intends to build 1,000 Starships over 10 years or so, then his engine output will be nearly identical to Continental or Lycoming or Pratt & Whitney. At their production volumes, very little cost savings from mass manufacturing is possible. Continental and Lycoming have cut their staffs to the bone and spent enormous amounts of money and effort to streamline or automate their production facilities, but their engine prices remain almost unchanged from what they were 10 years ago before those processes were in full swing (driven by the downturn in the economy). In terms of relative cost, inflation has lowered the prices of their products relative to the amount of money they have to spend, but the true cost to produce them (labor and materials, but mostly labor) hasn't budged.

Instead of the Lycoming IO-720's outlandish price for its performance, let's consider the economics of two far more common engines in experimental amateur built and general aviation:

GM's Corvair engine can be converted to an aircraft engine using aftermarket parts. There were millions of them made many decades ago, yet the cost to build one into an aircraft engine hovers around $15K for an honest 100hp aircraft engine. Continental's O-200 purpose-built 100hp light aircraft engine costs $22K brand new from the factory. Apart from the even more expensive lightweight O-200D models, both engines weigh nearly the same when all accessories are mounted and they're actually ready to fly. Both use aluminum crank case halves, oil pans, cylinder heads, and pistons with steel cylinder barrels / crankshafts / camshafts / pushrods / rocker arms (2 overhead valves per cylinder for both engines) and various other little bits that are incredibly similar. They wouldn't look out of place on a car or tractor from the 1930's. The Corvair actually has more parts (6 cylinders for the Corvair vs 4 for the Continental), but the parts are a little cheaper to come by because GM made nearly 4 million of them vs around 100,000 or so for the O-200. GM only made Corvair engines for a decade or so, but Continental's O-200 has been in production since 1947. I guess I also forgot to mention that most of the extra $7K was paying for specialized fabrication and testing labor, lots of inspection and testing equipment, and maintaining the production line to produce more components for the engine for longer than my father has been alive.

What was the point of telling you all of that?

It's not to regale you with stories from the aircraft engine world. There's a lesson on how much mass manufacturing actually lowers the marginal cost of complex and/or finely-crafted machines in there. The next time someone comes along and proclaims, "I'm going to make a 100hp aircraft engine for five grand!", you and others should respond with, "That sounds fantastic, but show me your production methods, testing methods, and what materials you're using."

I'm building my own Continental O-300 (6 cylinder version of the O-200) for my own aircraft using parts I've scrounged from various sources. The engine will run me about $15K by the time I'm done and I'm providing the labor for assembly and testing. It's dry weight is around 270 pounds. The dry weight of the Pratt & Whitney PT-6A is also around 270 pounds without the exhaust, but both are over 300 pounds when you add everything that has to be added to fly. I think my engine will produce around 150hp. I've seen 750hp PT-6A's go for around $500K used and in need of overhaul. FYI, if you see any aircraft engine listed for an absurdly low price, there's probably a good reason. Granted, that turboprop is at least 5 times more powerful than my engine for a given weight, but the PT-6A didn't cost 5 times more than my O-300 for that extra horsepower- it cost 25 times more! It'd be 50 times more if it was brand new from the factory and airworthy.

Do you have any idea how silly it sounds when ANYONE (by that I mean you, Elon Musk, or anyone else for that matter) claims that they're going to make a 400,000lbf rocket engine for far less than the cost of a PT-6A, but using turbo pumps that are wildly more powerful than the PT-6A that operate in wildly more severe conditions than a PT-6A?

I totally understand the hero worship. What I don't understand is the split from reality on basic economics. I don't care how much I personally admire someone because of what they do, I'm not going to believe them when they claim they're going to make something for 100 times or more less than what it typically costs, even if they're already producing something that's marginally less costly than an industry average. Those kinds of statements are just nonsense and you should know that.

So the main reasons for Starship being much cheaper are:

1. It's going to be a much simpler beast - a two stage rocket not a 3 rockets strapped together affair.

If that were true, then Falcon 9 should be significantly cheaper to operate than it is and simpler than Starship for sure if your line of argumentation held water. However, I think most of the costs of operating a reusable rocket aren't directly related to flying it.

Starship is far more complex than Falcon Heavy, especially the Raptor engines themselves. Adding ten extra and substantially more powerful engines to the booster and five extra and substantially more powerful engines to the upper stage doesn't reduce complexity. Staged combustion is also more complicated than gas generator to run reliably. Granted, that's not saying much since both types of engines are complex. We've reduced the number of staging events from 2 to 1, but that was the extent of the cost savings through design and operations simplification. That alone didn't make the rocket significantly less expensive.

2. It's going to be fully reusable.

This certainly helps, just not as much as many people seem to think. I've yet to see an orbital class vehicle go through reentry without significant refurbishment. I guess we'll know how much of a problem reentry heating is soon enough.

3. Because so many Starships are going to be produced, economies of scale kick in and the cost per unit falls hugely.

I've already illustrated how making 4 million engines made over a decade vs a hundred thousand engines made over the better part of a century affected cost. There simply wasn't a night-and-day difference in marginal unit cost. I can come up with many more examples from mass vs specialty engine production if that wasn't sufficient to prove the point. If the cost reduction argument doesn't work for a car engine, it won't work for a jet or rocket engine, either.

At best, if all 1,000 Starship propellant tanks / hulls were made over the course of a few years there would be cost savings associated with bulk purchasing of materials and not maintaining a standing army of aerospace welders for longer than necessary. The big question then is, where would you store all of them if a stiff wind can knock them over when they're empty?

4. Because there will be so many flights the costs per flight are going to be much smaller.

SpaceX already has a significant share of the market for launch services and the demand for launch services didn't substantially increase just because SpaceX provided a cheaper solution. It's almost as if there's a limit on demand that doesn't care how cheap the launches are. We need to figure out how to market this kind of stuff to the average consumer and to corporations to increase demand.

louis · 2020-03-07 18:25:52

Paul Wooster, the Principal Mars Development Engineer for Space X has stated 5 launches should be sufficient to refuel a Starship in LEO.
He said "more or less refuel" in one of his talks, so maybe they don't need 100% refuelling.

https://arstechnica.com/science/2019/06 … fely-back/

"An estimated five Starship launches are required to refuel a single Mars-bound Starship in low-Earth orbit, and this would involve the transfer of hundreds of tons of methane and liquid oxygen. "

I am myself a bit sceptical about the 100 tons to Mars, given they have already had a major revision down to 100 from 150 tons. But even if it was to reduce to 80, that would still be 400 tons overall (with 5 Starships landing on Mars) - enough to secure Mission One in my view.

GW Johnson wrote:

Well, look at this another way: are we doing better now than we were before?
Apollo era: $30B in 1970's $ to land 12 men on the moon. Allowing for inflation that's close to $100B in today's $. There were 7 Mercury flights, about a dozen Gemini flights, and about a dozen Saturn flights counting the tests. That's around $100B/2020$ for some 30 flights, or about $3B per launch.
The verified costs of the Space Shuttle were also right at $1B per launch in 1990 $. In today's $, that's probably no more than $2B/launch, because inflation hasn't been so severe in recent decades.
NASA with SLS/Orion is doing what it did during Apollo with Shuttle tinkertoys. They have yet to fly after two decades development, but the projected price for doing it the old way is $1-2B/launch in today's $. Why should that be surprising at all? The initial version of SLS is said to be capable of delivering 70 tons to LEO. That's roughly $21M/ton to LEO, give or take a big uncertainty.
ULA no longer flies the Delta-4 Heavy. They never made any attempt to reduce launch costs with it. As a result, it only launched military items in the last years it was used, at something like $15-20M/ton, about like NASA.
ULA did attempt to reduce costs with its Atlas-5, in order to compete with Spacex. Depending on which version, how many engines in the Centaur stage, and how many SRB's, their price per launch is in the $100-130M/launch range in today's $. Order of magnitude lower than NASA's track record, but a lower payload, too. About 20 metric tons to LEO. Call it $5-6M/ton to LEO. Yep, they're doing it a lot better than NASA ever did, by a factor of 3 or 4!
Spacex has been driving the competition in the commercial launch market, which is the motivation for lower launch costs. Falcon-9 now delivers almost 20 tons max to LEO (closer to 10 if you recover the booster) for a published price of $63M/launch not-recoverable. I haven't seen anything on the price break for recoverable-booster flights; take a guess and call it $50M. You're looking at something in the $3-5M/ton range for LEO. That's even better than ULA, and way to hell-and-gone better than NASA.
Spacex's Falcon-Heavy has only flown twice, but is listed on the website at around $85M per launch, and 63 tons to LEO, flown nonrecoverable. That's around $1.3-1.4M/ton to LEO. Remarkable! Yep, size matters: the bigger ship carrying more cargo really is cheaper to build and operate, even if it's a spaceship, not an ocean-going ship.
This stuff is not a linear curve, it's a curve of diminishing returns. I would expect the Starship/Super Heavy to have an even lower price per ton to LEO than Falcon Heavy, just because it is so much bigger than Falcon Heavy. Not dramatically lower (orders of magnitude, that's nonsense), but maybe around $0.5-1M/ton to LEO, once they get past all the teething troubles they are going to have with it.
I ran some numbers on this Starship/Super Heavy system, based on the various presentations that have been made public by Musk. I think it takes 6 not 5 tanker flights to fully refill a Starship on-orbit in LEO. And you don't go anywhere useful without a full refill. Not the moon, not Mars, not anywhere else.
Now I am likely wrong, but on the overoptimistic side. Spacex has yet to come anywhere near the vehicle inert mass they were projecting when I made those estimates. It might be a lot more tankers than 6 to refill, and I'd bet real money the payload to Mars will struggle to ever be 100 tons.
They think they're having teething troubles with tank welds now -- wait till they actually fly and start crashing or exploding vehicles. This spaceflight thing is not for the faint-of-heart or the low-of-budget, even when doing it orders and orders of magnitude cheaper than NASA.
Sorry to rain on anybody's parade, but those are my best projected guesses, based on 2020 $.
GW

louis · 2020-03-07 18:43:49

I think the many-multiple flights are for the flights to LEO. Interplanetary Starships won't be flying that often, indeed.

If Musk is planning on ferrying 1 million tons to Mars, that's 10,000 interplanetary flights which would require 50,000 flights but the fleet is only meant to be 1000 Starships, so I think he must be envisaging a much large number than 10 for multiple flights.

I don't know how good the construction method is but presumably the smaller the sections, the easier it is to assemble on a production line. Eventually I would presume robot welders would be used:

https://www.youtube.com/watch?v=HUU3HdxOqZs

SpaceNut wrote:

Replies for Louis, reading comments first then article...
Since sheets of stainless would be easier to use for shaping a ship it sounds to me like the use of Barrels is a recycling of materials which are junk value. No wonder its costs are under the Falcon 9 or of its heavy....Now if you are using the "barrel" as a term for the cylinder shaped sections of the outer shell then they must be using a lot of cheap stainless not from the US.
reading from the photo, which is the fuel tank construction image where the term "barrel" is used is speaking of sections of the construction assembly. The dome is made from shaped cut pieces that are welded together which accounts for failure along the seams. Little wonder why as the number goes up so will the rate of failure. Normal tanks are a single unit constructed tested and then placed into an assembly where its going to be used.
Trying to cheat material mass with this construction technique of intergrated piece assembly increase risk for failure of the tank as it can not be proven to be good as the shell is part of its assembly rather than a seperate part of the structure of the ship. Its also the reason for needing more man hours to build in a shorter period of time for that large piece cutting, shaping and welding build.
A starship can not fly 100 flights as the trips to mars ( 2yr approximate) are not an up and down trip with not to meantion heatshielding wear will mean replacing if you do keep the majority of the vehicle.
The first stage BFR will not do 100 flights as well since we have predictable falcon cores with a reuse rate of just 10 not counting landing errors. So no 100 flights with these either.

louis · 2020-03-07 19:06:25

I can't comment on your experience with purchasing aircraft parts. My own limited experience in industry is that bigger operations do result in very significant economies of scale.

I'm not engaged in hero worship! Musk might have been drawing on the spliff too strongly for all I know. But I think his comments do have to be treated with some respect since he has a pretty good track record at making things work which people said couldn't work.

That said, while researching this topic I found references to $2 million costs per flight of the Starship - $900K fuel and $1.1 million "operational costs" - this was from an earlier interview with Musk:

https://techcrunch.com/2019/11/06/elon- … er-launch/

So I am now wondering whether the Ars Technica journo has maybe misinterpreted and misquoted Musk - confusing flight cost with build cost?

If the flight cost is $2 million well that would give you $12 million for launch and refuelling, to get you to Mars. That would be $120 per Kg to Mars. No doubt there would be some additional costs associated with the Mars end of things and coms with ground control.

kbd512 wrote:

Louis,
I have worked in supply chain for consumer products for a number of different major manufacturers, all of them global operations, since 2007. They were involved in the manufacture of everything from foodstuffs, beverages, pharmaceuticals, sports apparel and equipment, computers and other consumer electronics, various kinds of steel or steel parts, and now for an oilfield services company that makes its own tools and purchases mass quantities of materials to drill for oil and natural gas, both here in America and around the world. That's why I know a little bit about what it costs to get something from Point A to Point B, at least here on Earth. Everyone I've worked for is moving ships and rail cars and trucks filled with raw materials and finished good products.
In my personal life, I have looked high and low for that true bargain of the aircraft engine world. All I can tell you about that is that there aren't many bargains to be had. Anything manufactured in the low thousands using highly specialized machining methods, certified materials, multiple material and finished part inspections, and lots of touch labor is inherently expensive. All aircraft engines, which includes rocket engines, will remain very high dollar items as a function of low production volumes and high labor costs unless something breaks that equation.
Fabricating 1,000 vehicles with 42,000 engines does zip / zero / zilch / nada to break that equation. If Elon Musk truly intends to build 1,000 Starships over 10 years or so, then his engine output will be nearly identical to Continental or Lycoming or Pratt & Whitney. At their production volumes, very little cost savings from mass manufacturing is possible. Continental and Lycoming have cut their staffs to the bone and spent enormous amounts of money and effort to streamline or automate their production facilities, but their engine prices remain almost unchanged from what they were 10 years ago before those processes were in full swing (driven by the downturn in the economy). In terms of relative cost, inflation has lowered the prices of their products relative to the amount of money they have to spend, but the true cost to produce them (labor and materials, but mostly labor) hasn't budged.
Instead of the Lycoming IO-720's outlandish price for its performance, let's consider the economics of two far more common engines in experimental amateur built and general aviation:
GM's Corvair engine can be converted to an aircraft engine using aftermarket parts. There were millions of them made many decades ago, yet the cost to build one into an aircraft engine hovers around $15K for an honest 100hp aircraft engine. Continental's O-200 purpose-built 100hp light aircraft engine costs $22K brand new from the factory. Apart from the even more expensive lightweight O-200D models, both engines weigh nearly the same when all accessories are mounted and they're actually ready to fly. Both use aluminum crank case halves, oil pans, cylinder heads, and pistons with steel cylinder barrels / crankshafts / camshafts / pushrods / rocker arms (2 overhead valves per cylinder for both engines) and various other little bits that are incredibly similar. They wouldn't look out of place on a car or tractor from the 1930's. The Corvair actually has more parts (6 cylinders for the Corvair vs 4 for the Continental), but the parts are a little cheaper to come by because GM made nearly 4 million of them vs around 100,000 or so for the O-200. GM only made Corvair engines for a decade or so, but Continental's O-200 has been in production since 1947. I guess I also forgot to mention that most of the extra $7K was paying for specialized fabrication and testing labor, lots of inspection and testing equipment, and maintaining the production line to produce more components for the engine for longer than my father has been alive.
What was the point of telling you all of that?
It's not to regale you with stories from the aircraft engine world. There's a lesson on how much mass manufacturing actually lowers the marginal cost of complex and/or finely-crafted machines in there. The next time someone comes along and proclaims, "I'm going to make a 100hp aircraft engine for five grand!", you and others should respond with, "That sounds fantastic, but show me your production methods, testing methods, and what materials you're using."
I'm building my own Continental O-300 (6 cylinder version of the O-200) for my own aircraft using parts I've scrounged from various sources. The engine will run me about $15K by the time I'm done and I'm providing the labor for assembly and testing. It's dry weight is around 270 pounds. The dry weight of the Pratt & Whitney PT-6A is also around 270 pounds without the exhaust, but both are over 300 pounds when you add everything that has to be added to fly. I think my engine will produce around 150hp. I've seen 750hp PT-6A's go for around $500K used and in need of overhaul. FYI, if you see any aircraft engine listed for an absurdly low price, there's probably a good reason. Granted, that turboprop is at least 5 times more powerful than my engine for a given weight, but the PT-6A didn't cost 5 times more than my O-300 for that extra horsepower- it cost 25 times more! It'd be 50 times more if it was brand new from the factory and airworthy.
Do you have any idea how silly it sounds when ANYONE (by that I mean you, Elon Musk, or anyone else for that matter) claims that they're going to make a 400,000lbf rocket engine for far less than the cost of a PT-6A, but using turbo pumps that are wildly more powerful than the PT-6A that operate in wildly more severe conditions than a PT-6A?
I totally understand the hero worship. What I don't understand is the split from reality on basic economics. I don't care how much I personally admire someone because of what they do, I'm not going to believe them when they claim they're going to make something for 100 times or more less than what it typically costs, even if they're already producing something that's marginally less costly than an industry average. Those kinds of statements are just nonsense and you should know that.

So the main reasons for Starship being much cheaper are:
1. It's going to be a much simpler beast - a two stage rocket not a 3 rockets strapped together affair.
If that were true, then Falcon 9 should be significantly cheaper to operate than it is and simpler than Starship for sure if your line of argumentation held water. However, I think most of the costs of operating a reusable rocket aren't directly related to flying it.
Starship is far more complex than Falcon Heavy, especially the Raptor engines themselves. Adding ten extra and substantially more powerful engines to the booster and five extra and substantially more powerful engines to the upper stage doesn't reduce complexity. Staged combustion is also more complicated than gas generator to run reliably. Granted, that's not saying much since both types of engines are complex. We've reduced the number of staging events from 2 to 1, but that was the extent of the cost savings through design and operations simplification. That alone didn't make the rocket significantly less expensive.
2. It's going to be fully reusable.
This certainly helps, just not as much as many people seem to think. I've yet to see an orbital class vehicle go through reentry without significant refurbishment. I guess we'll know how much of a problem reentry heating is soon enough.
3. Because so many Starships are going to be produced, economies of scale kick in and the cost per unit falls hugely.
I've already illustrated how making 4 million engines made over a decade vs a hundred thousand engines made over the better part of a century affected cost. There simply wasn't a night-and-day difference in marginal unit cost. I can come up with many more examples from mass vs specialty engine production if that wasn't sufficient to prove the point. If the cost reduction argument doesn't work for a car engine, it won't work for a jet or rocket engine, either.
At best, if all 1,000 Starship propellant tanks / hulls were made over the course of a few years there would be cost savings associated with bulk purchasing of materials and not maintaining a standing army of aerospace welders for longer than necessary. The big question then is, where would you store all of them if a stiff wind can knock them over when they're empty?
4. Because there will be so many flights the costs per flight are going to be much smaller.
SpaceX already has a significant share of the market for launch services and the demand for launch services didn't substantially increase just because SpaceX provided a cheaper solution. It's almost as if there's a limit on demand that doesn't care how cheap the launches are. We need to figure out how to market this kind of stuff to the average consumer and to corporations to increase demand.

SpaceNut · 2020-03-07 19:28:02

The numbers do not add up or even come close to working for the bfr starship tankers to the bfr launch 100 t cargo and are not any better for the crewed combination either. So if we are landing 5 ships for cargo and 1 for a crew in each window of launch after filling up with fuel by the 5 to 6 tankers for each unit we are in trouble as the boiloff for each bfr for the units mean all refueling must happen at the same time. So 6 ships sent to orbit followed by 36 more to fill them up before they can leave to get to mars....Once they arrive thats 2 years more to be able to repeat the same sequence.

louis · 2020-03-07 20:15:37

As indicated, Space X's Principal Engineer says 5 flights for refuelling one Starship. And I think the Mission One proposal is for 5 Starships in total. But they don't all have to head off for Mars at exactly the same time. There's a two month window for transit to Mars.

SpaceNut wrote:

The numbers do not add up or even come close to working for the bfr starship tankers to the bfr launch 100 t cargo and are not any better for the crewed combination either. So if we are landing 5 ships for cargo and 1 for a crew in each window of launch after filling up with fuel by the 5 to 6 tankers for each unit we are in trouble as the boiloff for each bfr for the units mean all refueling must happen at the same time. So 6 ships sent to orbit followed by 36 more to fill them up before they can leave to get to mars....Once they arrive thats 2 years more to be able to repeat the same sequence.

kbd512 · 2020-03-07 20:27:22

Louis,

I've no doubt that making a production run of 1,000 rockets is more economical than producing them one at a time. However, even if SpaceX is making 1,000 rockets and 42,000 engines, no night-and-day economies of scale are possible. If we were talking about setting up a production plant like Tonawanda to produce 1,000,000+ rocket engines, then we might see a more significant portion of that economy of scale come to fruition. However, no flying machine in the history of flight has exceeded the production run of the lowly Cessna 172 that I fly and those now cost half a million dollars, despite having the same engines in them that they did when my father was a kid, being fabricated of the same materials, and only having minor changes to the avionics over the years. The only thing that significantly changed was the cost of labor and the materials costs actually went down, as a percentage of the total fabrication costs. Furthermore, all other aircraft of similar configuration and performance have seen similar cost increases, again, without anything else apart from labor costs drastically changing over the years. When my father was a kid, a 4 seat aircraft like the Cessna 172 was equivalent in price to 2 Cadillacs or so. Today, it's anywhere between 10 and 14 Cadillacs. It's the same airframe, powered by the exact same engine, and uses the same materials.

There has been endless experimentation with methods to make lower-cost flying machines, but they're still far more expensive to make than cars, precisely because of the labor rather than material costs. Nearly every vehicle on the road has substantially more material in it than an aircraft with equivalent seating capacity, but you have to attain super car performance levels before you approach the cost of the simplest of aircraft with similar speed performance. With nearly identical production volumes, we shouldn't expect other aerospace vehicles such as rockets to become fantastically cheaper, irrespective of any wild claims utterly lacking for evidence. Whenever SpaceX or other aerospace companies start making outlandish claims, they're usually aspirational goals that are real short on evidence to indicate how it is that they might achieve those goals.

Here's what I think:

1. Achieving higher production rates and using cheaper materials will lower production costs. There's no doubt that Aluminum, Titanium, and Carbon composites have wonderful qualities for aerospace vehicles, but all drastic increases in performance typically carry rather stiff cost and operating limitation penalties with them. GW has pointed out numerous times that while Titanium is lighter than steel and may not oxidize as readily as steel at elevated temperatures, most of its strength-to-weight advantage goes right out the window when it operates at the same temperatures that would require stainless steel.

I could imagine a fabrication cost somewhat better than that of an airliner using lower cost materials and fewer unique parts, but the requirement for 42 engines is wildly different from any commercial airliner, which only has 2 or 4 engines that require maintenance. Despite far lower performance requirements, those aircraft engines are hideously expensive to build and maintain. A rocket engine is just another gas turbine application with performance requirements far in excess of a jet engine. I fail to see how that performance level will be attained for far less cost, though I'd love to be proven wrong.

2. Rockets are not disproportionately affected by the marginal increase in weight from using heavier but more heat and fatigue resistant materials such as steels, on account of how brief their flights are and the fact that the weight advantages of lighter materials are at least partially offset by increased heat shield mass. Gravity only acts on them for a few minutes and they primarily use thrust rather than aerodynamic lift to fly. I can totally buy SpaceX's and Elon's argument that stainless steel is a more practical structural material choice than Aluminum or Titanium or CFRP for rockets, given their operating environments. This argument has been made in the past and evidence from actual testing was even provided which indicated that the use of low cost and high-strength alloy steels that were easy to weld could significantly reduce the fabrication costs of simplistic rockets. There was sub-scale testing of the "big dumb booster" concept that used improved thrust to overcome its higher inert weight and the OTRAG series of rockets. Making lots of thrust is the name of the game and Raptor provides that in spades.

3. Reliable rocketry has never been a particularly economical endeavor due to the amount of hardware and fuel and R&D required to produce enough speed to attain orbital velocity. The only night-and-day difference in rocketry will come from eliminating the need for a booster while maintaining the payload performance of the upper stage. There's no reason to think that we're going to get orders-of-magnitude improvements in cost merely by switching to lower cost fabrication materials and making the rockets reusable, unless we can also do that with little to no refurbishment prior to re-flight.

GW Johnson · 2020-03-07 21:19:02

We'll see about 5 vs 6 vs 7 tankers to refill a Starship on-orbit in LEO. As near as I can tell, they have not yet come anywhere near the projected inert mass fraction. Inert + propellant + payload mass fractions MUST sum to 1. Anything else is deluded fantasy.

Not fully refilling cuts mass ratio which cuts delta-vee. Not a smart thing to do, when flying to the moon or Mars. At the very least, it cuts your payload. If worse, you can't get there from here. Dead crew.

GW

kbd512 · 2020-03-08 01:23:33

There's a rather interesting piece of errata from aircraft design history that lends creedence to the argument that steel is a better material choice for airframes when ultimate durability and fabrication cost are important (as they generally tend to be for an owner-operator). The Meyers 200D used a substantial tubular steel support structure that made the airframe very strong and durable, to the point that it remains one of the few or possibly the only general aviation airframe that's never had an AD (Airworthiness Directive) issued by the FAA against it due to an airframe fatigue structural failure. It was a product of the 1950's and like so many other innovations from that era, there are still a few flying around today. A variety of Russian trainer aircraft were similarly well-built to withstand the punishment that novice pilots would inevitably dish out. If you've ever seen that airframe without the skin attached, it looks a lot like a NASCAR or drag racing vehicle chassis. The flap system also gives an unusually high (for piston-engined GA aircraft, not most business jets) trimmed maximum lift coefficient (typically determines stall speeds associated with slowing down to land), so it's every bit as fast in cruise as the most impressive composite airframes designed by Burt Rutan and fuel burn is not significantly better or worse than any other aircraft that can cruise as fast as it does.

GW Johnson · 2020-03-08 09:55:10

Steel as airframe structure makes a lot of sense when it is not the bulk of the airframe. That's why covered steel tube airframes were so durable while still being light enough to serve. Piper Cubs were built the same way Kbd512 describes, just without flaps, having a bigger wing instead.

You use steel as skin material only if you are forced to, because of the higher weight. Usually it is heat resistance from flying very fast that forces this choice. Titanium is sort of intermediate between steel and aluminum in terms of heat resistance. Where it is adequate thermally, it is half the weight of steel. Where it is not, you must use the steel. And by steel, I mean the stainlesses (the austenitic 300-series, and the martensitic alloys like 4130, etc, not plain low-carbon steel, which has the same thermal limits as titanium).

Depending upon your peak heating rate, it may be just barely possible to have exposed stainless skins without any heat shield, in a vehicle with a very low ballistic coefficient, at 8 km/s LEO entry speeds. Most of the time, it will not be possible to design to a ballistic coefficient that low, so you will need a suitable substrate under a heat shield, on all windward skin surfaces.

You will NEVER get exposed metal leading edges and nosetips to survive without heat shielding of some kind. That heating rate problem is about an order of magnitude more severe.

Speed up the entry interface speed, and it's heat shields all the way around, even on leeward skins. 11 km/s returning from the moon. 12-17 km/s returning from Mars. Only arrival at Mars is similar to LEO entry at ~7.5 km/s. From Mars orbit, it is only 3.6 km/s.

Heat shields: there are ablatives, and there are refractories. The ablatives are things like PICA-X, which if thick enough (making them heavier) can survive LEO entry more than once. Maybe 4 times or thereabouts, at practical thicknesses. You have to be sure they stay glued or attached in some way to the substrate, and they are largely their own heat sink for the thermal wave headed inward after entry.

The refractories have to be low density, or they cannot work more than once (which rules out all the new "super ceramics", they are all very high density). The prime examples are the low density tiles and thermal blankets used on the Space Shuttle, and now on the X-37B. They are too lightweight to serve as their own heat sink for the thermal wave; the substrate must do that. If there is not too big a thermal wave percolating inward, aluminum can serve. If there is too much heat to sink, you must use titanium, or even steel.

You cannot use this low density ceramic stuff all the way to its melt point, because of a solid-solid phase change that induces shrinkage and severe cracking. That is fundamentally why it does not protect nose tips and leading edges: they get too hot for it. They just barely serve as windward-side lateral skin protection at LEO entry speeds (8 km/s). Faster is just NOT feasible with any material we know of.

If you were designing an orbit-to-orbit Mars transport, only the landers and emergency escape capsules would need to survive entries, and those would be at 3.6 km/s at Mars, and 8 km/s at Earth. You could use low-density ceramics, with a bit of carbon-carbon on any nosetips and leading edges at Earth.

If you design around direct entries from the interplanetary trajectories, you are looking at 7.5 km/s entry speed at Mars, but some 12-17 km/s entry speed at Earth return. The only feasible heat shield material is ablative. The best known is PICA-X. Period.

See how mission architecture drives thermal survival, which drives your vehicle design?

GW

Last edited by GW Johnson (2020-03-08 10:03:19)

louis · 2020-03-08 16:10:18

Here's a relevant discussion:

https://www.reddit.com/r/SpaceXLounge/c … eavy_cost/

Some people suggesting a cost of $50 million minimum for a basic cargo Starship-Superheavy stack.

If you can develop full resuability and get say 100 flights (with periodic refurb at, maybe, let's guess $20million) that would give you a build cost of $700,000 per flight.

louis · 2020-03-09 11:40:28

I'm coming back to the view that Musk's claim of $5 million per ship might not be so insane.

This is Robert Zubrin's take on the cost of a Starship:

"If Musk set up a similar line with a workforce of 3,000, that would mean labor costs on the order of $6 million per ship, or between $15 to $20 million each, with materials and avionics included."

This is based on a built rate of 50 per annum.

https://www.nationalreview.com/2020/02/ … G8lDdP7RbM

louis · 2020-03-09 18:01:41

An interview with Elon Musk shedding some more light on some of the issues covered above:

https://www.youtube.com/watch?v=UFlviuwr2jg

SpaceNut · 2020-03-09 18:38:47

Timeline of indecision....
Dec 12, 2018 · News SpaceX’s BFR program pursuing advanced Starship heat shield with NASA help BFS/Starship shows off some of its heat shield. SpaceX may be looking into an advanced NASA solution for BFR

https://www.teslarati.com/spacex-new-bf … ssistance/
Musk has given spacefaring fans a glimpse of the hexagonal heatshield tiles that will eventually protect the craft from searing heat.

According to a Space Act Agreement signed by SpaceX and NASA’s Ames Research Center in June 2018, the private company has begun working with NASA to acquire some basic experience and lessons-learned with a thermal protection (heat shield) material that is largely new to SpaceX. Known as TUFROC (short for Toughened Uni-piece Fibrous Reinforced Oxidation-Resistant Composite), the NASA Ames-developed material is capable of withstanding temperatures as high as 1700 C (~3100 F) and is apparently an item of interest to SpaceX’s next-gen BFR (Starship/Super Heavy) rocket program, particularly Starship’s heat shield.

Thats a little higher than what the shuttle saw on its reentry for the tiles that it had.

Jan 25, 2019 · Starship’s stainless steel heat shield will be peppered with tiny holes to allow super-cold liquids, like liquid methane to pass through it.

Mar 18, 2019 · SpaceX tests heat shields that will stop its Starship from burning up "White-hot parts reached orbital entry temp of around 1,650 Kelvin,"

Musk tweeted that the tiles are hexagonal-shaped because that provides "no straight path for hot gas to accelerate through the gaps." The tiles will be installed on the windward side, towards the direction of re-entry, "with no shielded need on the leeward side." The hottest sections will have a "transpiration cooling" system, with microscope pores on the exterior that allow water or methane to ooze out and cool the exterior. That would minimize damage on the heat shields and allow the Starship to return to service shortly after a flight merely by refilling the heatshield reservoir. "Transpiration cooling will be added wherever we see erosion of the shield,".

Which means he does not know whether he needs it or not.....

GW Johnson · 2020-03-10 11:10:08

Active cooling is either heavier than passive cooling, or it is very leaky. Or both.

Active cooling is usually a double wall in one way or another. That is inherently heavier by a factor of 2 to 4 for any sort of low-pressure tank or atmosphere-filled space. There is NO way around that.

If you stick with a single-wall vessel that contains a liquid, you can use porosity to allow the liquid to leak through to the exterior, as a sort of transpiration or evaporative cooling agent. But that means your tank is leaky, even when you fill it on the launch pad. You cannot have it both ways. If it leaks to cool during entry, it leaks, period, all the time.

And, you cannot do the leaky-wall approach for an atmosphere-filled space. You MUST do the double-wall thing. Have to transport the liquid to where it is needed. Heavy.

Passive cooling is a poor choice of words. The right choice is radiation cooling. If the surroundings are transparent to translucent, then a hot surface can thermally radiate to the far surroundings. If you are adjacent to Earth, those far surroundings are 300 K Earth temperatures, not the 3-4 K of deep space.

How much power you can radiate depends upon how hot your surface can be (fourth power temperature), and upon how efficient your surface emissivity is at the color temperature at which you radiate. It is linearly proportional to emissivity, which can vary between 0 and 1, and usually varies strongly with wavelength.

What you need for reentry is radiative emission equal to the sum of convective and radiative heating, at the peak heating rate during entry. This has to happen at a surface temperature you can tolerate, and the plasma in the boundary layer has to be at least translucent, for your skin radiation to get through.

That's just physics, guys!

GW

Last edited by GW Johnson (2020-03-10 11:12:53)

louis · 2020-03-10 15:34:30

What about spraying the key areas with the cooling liquid? Wouldn't that cut down on additional mass? Just a thought!

GW Johnson wrote:

Active cooling is either heavier than passive cooling, or it is very leaky. Or both.
Active cooling is usually a double wall in one way or another. That is inherently heavier by a factor of 2 to 4 for any sort of low-pressure tank or atmosphere-filled space. There is NO way around that.
If you stick with a single-wall vessel that contains a liquid, you can use porosity to allow the liquid to leak through to the exterior, as a sort of transpiration or evaporative cooling agent. But that means your tank is leaky, even when you fill it on the launch pad. You cannot have it both ways. If it leaks to cool during entry, it leaks, period, all the time.
And, you cannot do the leaky-wall approach for an atmosphere-filled space. You MUST do the double-wall thing. Have to transport the liquid to where it is needed. Heavy.
Passive cooling is a poor choice of words. The right choice is radiation cooling. If the surroundings are transparent to translucent, then a hot surface can thermally radiate to the far surroundings. If you are adjacent to Earth, those far surroundings are 300 K Earth temperatures, not the 3-4 K of deep space.
How much power you can radiate depends upon how hot your surface can be (fourth power temperature), and upon how efficient your surface emissivity is at the color temperature at which you radiate. It is linearly proportional to emissivity, which can vary between 0 and 1, and usually varies strongly with wavelength.
What you need for reentry is radiative emission equal to the sum of convective and radiative heating, at the peak heating rate during entry. This has to happen at a surface temperature you can tolerate, and the plasma in the boundary layer has to be at least translucent, for your skin radiation to get through.
That's just physics, guys!
GW

louis · 2020-03-10 15:49:55

Here's the full version of the Musk video re Satellite 2020:

https://www.youtube.com/watch?time_cont … =emb_title

Some interesting thoughts as always:

"Technology does not automatically improve."

He's obviously concerned that a Moon or Mars base won't happen without more innovation. He emphasises the crucial importance of software in all technologies now.

Maybe he's thinking about how to produce a fully automated propellant production facility? Or maybe it's the PV power system that is worrying him. Something is!

SpaceNut · 2020-03-10 18:13:04

Long and short pressure builds under a tile that is hot it will cause it to crack and or become no longer adhered to its location on the hull of the ship.

New Mars Forums

Announcement

#1 2020-03-06 19:35:50

An important article...

#2 2020-03-07 01:30:26

Re: An important article...

#3 2020-03-07 06:06:42

Re: An important article...

#4 2020-03-07 07:13:16

Re: An important article...

#5 2020-03-07 09:47:21

Re: An important article...

#6 2020-03-07 09:52:12

Re: An important article...

#7 2020-03-07 10:18:57

Re: An important article...

#8 2020-03-07 16:36:43

Re: An important article...

#9 2020-03-07 18:25:52

Re: An important article...

#10 2020-03-07 18:43:49

Re: An important article...

#11 2020-03-07 19:06:25

Re: An important article...

#12 2020-03-07 19:28:02

Re: An important article...

#13 2020-03-07 20:15:37

Re: An important article...

#14 2020-03-07 20:27:22

Re: An important article...

#15 2020-03-07 21:19:02

Re: An important article...

#16 2020-03-08 01:23:33

Re: An important article...

#17 2020-03-08 09:55:10

Re: An important article...

#18 2020-03-08 16:10:18

Re: An important article...

#19 2020-03-09 11:40:28

Re: An important article...

#20 2020-03-09 18:01:41

Re: An important article...

#21 2020-03-09 18:38:47

Re: An important article...

#22 2020-03-10 11:10:08

Re: An important article...

#23 2020-03-10 15:34:30

Re: An important article...

#24 2020-03-10 15:49:55

Re: An important article...

#25 2020-03-10 18:13:04

Re: An important article...

Board footer