Sources of economic growth for SpaceX, reusable rockets

[Jim]

To expand on a thought regarding expansion of the number of launches for the commercial satellite market:
 from the 3% number Antonio gave, assuming this meant out of total expenditure which makes sense, it seems unlikely to me that reusability alone would completely change the business model of TV broadcast companies that would lead them to launch much more ofter. If we assume 100 - 200M$ for the satellite itself, which was a number frown here, this also confirms that there is more to it than just the rockets that keep the number of launches low. But what about the prospect of reusability coupled with innovation in satellite technology leading to a launch once in every 5 years instead of once in every 20 years or so?
Someone with more knowledge on the commercial satellite market might be more equipped to answer that.

There is a limited number of orbital slots for GSO comsats.  Spacecraft on similar frequencies can only be placed so close.  Also, what innovation?

[Elmar Moelzer]

It is funny. Antonio claimed that satellites will become smaller. This will actually be beneficial for SpaceX, because it will put more GSO satellites within the capabilities of the fully reusable F9. It would mean less work for the (probably less economic) Falcon Heavy but more work for F9.

[zd4]

To expand on a thought regarding expansion of the number of launches for the commercial satellite market:
 from the 3% number Antonio gave, assuming this meant out of total expenditure which makes sense, it seems unlikely to me that reusability alone would completely change the business model of TV broadcast companies that would lead them to launch much more ofter. If we assume 100 - 200M$ for the satellite itself, which was a number frown here, this also confirms that there is more to it than just the rockets that keep the number of launches low. But what about the prospect of reusability coupled with innovation in satellite technology leading to a launch once in every 5 years instead of once in every 20 years or so?
Someone with more knowledge on the commercial satellite market might be more equipped to answer that.

There is a limited number of orbital slots for GSO comsats.  Spacecraft on similar frequencies can only be placed so close.  Also, what innovation?

I was talking about commercial satellites more generally and not GSO specifically. But what about constellations in LEO?
Thinking in the direction of COMMStellation (http://en.wikipedia.org/wiki/COMMStellation), and what seems to be a general bloom in micro-satellites, what is the prospect of these taking replacing functionality on GSOs? I hope to not offend anyone with my ignorance on the subject, but - is there some overlap, at least theoretically, in what you can do with a large enough constellation and what you can do with a GSO satellite?

[sghill]

Thinking in the direction of COMMStellation (http://en.wikipedia.org/wiki/COMMStellation), and what seems to be a general bloom in micro-satellites, what is the prospect of these taking replacing functionality on GSOs? I hope to not offend anyone with my ignorance on the subject, but - is there some overlap, at least theoretically, in what you can do with a large enough constellation and what you can do with a GSO satellite?

There are two major differences between the two types that can't be engineered around through frequency hopping, timing windows within each band, hand-offs, signal combining, and other trickery because they involve physical limitations of the earth and where the satellite is located. 

The first is coverage and power needs to assure coverage.  An GSO sat can "see" more of the globe from its vantage point, but it's further out, so it needs more powerful receivers and transmitters- hence their large size and expense.  Lower orbit sats use less power, and are less complicated, but they cover a smaller portion of the globe, plus they move in relation to the globe, so you need several to equal the same coverage.

The second is distance, a lower satellite has a faster response time because the speed of light from the orbit to the ground isn't consequential.  This second limitation has largely gone away for two-way communications with the global network of fiber optics, but it still plays a role with satellite phones.  For one-way communications GSO is king.

I think your "hybrid" idea is workable with medium sized satellites in medium sized orbits, and a reusable booster may be perfect for getting them there.  Just thinking about back of the napkin constellation size, I think you'd need a constellation about the size of the 32 satellite GPS constellation, and not the 66 satellite LEO Iridium constellation to cover the earth, versus 3 or 4 GSO birds.

That's still only around a 30 satellite difference however, so not a lot of launches- especially if you send up more than one bird per launch- but nevertheless, several dozen launches is several dozen launches!

At any rate, it's fundamentally an accounting game, and not an engineering game.

[TrevorMonty]

There was talk of large constellation of 1600 small LEO satellites by Google to give global internet coverage. They definitely have money to make this happen.

[Llian Rhydderch]

SpaceX (and perhaps others) is going to have a reusable 1st stage booster (versus returnable) in the near future unless some technical difficulty proves that it's unworkable as a business model. But keep in mind that the development cost is going to be a sunk cost at that point. 

OK, so what do the economics of that reusable booster look like?  The US military isn't going to care much at first about reusability for its payloads because mission assurance plays an out-sized role in their launch cost calculations.  Non-compete launches won't be affected at all.  Civilian launches will care a great deal.  A million $$ is a million $$ to an investor, so even small improvements over launch costs by implementing reusability will wildly drive existing businesses to a reusable booster operator.  This assumes the MAJOR point that the benefits of reusability are ultimately passed along from the operator in the form of lower launch pricing to customers. They may not be.

Because existing business is very nearly a zero-sum game, commercial Soyuz, Arianespace, Orbital, and other commercial launch operators are going to take the hit first in the short term until they can respond with lower offered price points of their own through whatever means they have at their disposal.  A greater share of the existing market may be just fine for justifying the expense of maintaining a few reusable boosters (remember always that by this point they've already spent the money to develop it, so it's no longer counted) versus more cheaply building expendables.  And taking the thought to it's logical conclusion, reusability may ultimately eliminate any justification for expendables in a given booster class.  If reusability drives down launch costs to the operator or the customer (or both), expendability will go away for missions that use that sized booster.  There is no economic model where a customer buys a cargo ship to use it once- even if the value of the cargo is greater than that of the ship- when a reusable ship is available at a lower cost. 

Long-term this pressure will continue to trickle up into launches where competition exists, but where cost is secondary to mission assurance (such as U.S. military launches or NASA science launches).  That second phase may take years, allowing competitors time to respond, perhaps with reusability of their own, or perhaps with other pricing reduction techniques. 

Along with the zero-sum game of existing business however, is the positive-sum game of creating new markets with disruptive technology and techniques. 

If a reusable rocket (and lower launch pricing) can boost "dumb and dumber" items into space- such as construction materials instead of complex communications satellites- then a new market for frequent launches may develop as some entity seeks to make something up there or do anything other than the three primary space activities of launching self-contained telecommunications, science, and reconnaissance missions.  The list beyond these three things is endless if launch costs go down. If the dynamics of launch pricing change, then new customer entrants may need frequent, cheap, launch services for some different need, and that's where reusability could have its greatest impact.

To recap, here are the benefits of an operational reusable booster with (assumed) lower operational costs than CAPEX costs compared with an expendable booster:
1) internally for the operator- CAPEX per launch may be lowered through reusability to increase profits without changing market share (e.g. pricing doesn't change, but internal costs are lower).
2) existing market- reusability lowers launch costs and takes market share from other operators to make its business case.
3) new market- reusability lowers launch costs and creates new markets to make its business case.
Any combination or mash-up of the above works as well.

Recap of my recap:  If reusability drives down launch pricing, new launch purchasing opportunities will materialize along the demand curve- as would be expected with the pricing of any sort of widget.   If reusability drives down launch pricing and/or costs, expendability will go away for price-competition missions that use that sized booster.

This was a really good post, sghill.  Thanks for taking the time to set forth your argument, and your evidence for it, in such a comprehensive fashion.

[KelvinZero]

If NASA decided to pursue propellant depots for HSF I guess that would create a market a few times bigger than the current ISS cargo and crew, at least in terms of tonnage. Its an ideal application for a cheap reusable rocket needing repetitive missions.

[Llian Rhydderch]

To expand on a thought regarding expansion of the number of launches for the commercial satellite market:
 from the 3% number Antonio gave, assuming this meant out of total expenditure which makes sense, it seems unlikely to me that reusability alone would completely change the business model of TV broadcast companies that would lead them to launch much more ofter. If we assume 100 - 200M$ for the satellite itself, which was a number frown here, this also confirms that there is more to it than just the rockets that keep the number of launches low. But what about the prospect of reusability coupled with innovation in satellite technology leading to a launch once in every 5 years instead of once in every 20 years or so?
Someone with more knowledge on the commercial satellite market might be more equipped to answer that.

There is a limited number of orbital slots for GSO comsats.  Spacecraft on similar frequencies can only be placed so close.  Also, what innovation?

One example of potential innovation is the dramatic shrinkage and mass reduction of electronics, with knock-on effects in reduced mass for the power supply (solar) and the propellant mass and engine mass needed to keep the satellite in Geosynchronous orbit, and move it to a graveyard orbit at end-0f-life.  (Or, why not, maybe even future deorbit of the sat when the current unpriced externality regime of leaving derelict satellites in geocentric orbits eventually becomes a priced internal cost to the mission.  Reduced mass will be very important if this eventuality occurs).

The shrinkage in electronics and the entirely new server architectures made possible by CMOS semiconductor technology ate the lunch of the big mainframe and mini/microcomputer vendors like IBM, Sun Microsystems, et al.  It will happen in satellites too.

The availability of cheaper/faster/less massive technologies will allow substantial technology upgrades from the old, out-of-date, Rad-hardened silicon architectures used in previous generations of satellites and off-Earth probes.  While we don't know the lifetimes of these cheaper/better/faster-innovation technologies as applied to satellites, it is quite likely that a trade space is created to determine the economics of staying with 15-year nominal commsat life, or whether 7 or 5, or even 3, year turns might later on be the more economic option since the faster turns also allows quicker upgrades in available technologies. 

So, yes, not only is there substantial innovation on the horizon, there is a plausible argument that it could affect the Geosynch commsat market.

There are other examples as well.

[pippin]

If you look at the very long term perspective, GEO comsats will go away anyway. Communication in general is going towards IP/bidirectional communication and that's something not compatible with the link times of GEO sats. It's also not spectrally efficient enough due to the huge cell size.

So either we are going to see LEO constellations or most of the communication will become terrestrial again with cables as backbones. Probably the latter since LEO cells are still quite big.

In that kind of scenario GEO sats will only serve as cheap backbones for non time-critical data (batch link) and to cover broadcast comm to remote areas. The latter, too, might go away if LEO constellations would evolve.

So I doubt that in the very long term and assuming the kind of technical evolution you are describing the GEO Comsat market will see such a development towards cheaper, shorter, more.

[Jim]

One example of potential innovation is the dramatic shrinkage and mass reduction of electronics, with knock-on effects in reduced mass for the power supply (solar) and the propellant mass and engine mass needed to keep the satellite in Geosynchronous orbit, and move it to a graveyard orbit at end-0f-life.  (Or, why not, maybe even future deorbit of the sat when the current unpriced externality regime of leaving derelict satellites in geocentric orbits eventually becomes a priced internal cost to the mission.  Reduced mass will be very important if this eventuality occurs).

The shrinkage in electronics and the entirely new server architectures made possible by CMOS semiconductor technology ate the lunch of the big mainframe and mini/microcomputer vendors like IBM, Sun Microsystems, et al.  It will happen in satellites too.

The availability of cheaper/faster/less massive technologies will allow substantial technology upgrades from the old, out-of-date, Rad-hardened silicon architectures used in previous generations of satellites and off-Earth probes.  While we don't know the lifetimes of these cheaper/better/faster-innovation technologies as applied to satellites, it is quite likely that a trade space is created to determine the economics of staying with 15-year nominal commsat life, or whether 7 or 5, or even 3, year turns might later on be the more economic option since the faster turns also allows quicker upgrades in available technologies. 

  The electronics are a small part of the mass of a spacecraft, so your premise is wrong

[mlindner]

One example of potential innovation is the dramatic shrinkage and mass reduction of electronics, with knock-on effects in reduced mass for the power supply (solar) and the propellant mass and engine mass needed to keep the satellite in Geosynchronous orbit, and move it to a graveyard orbit at end-0f-life.  (Or, why not, maybe even future deorbit of the sat when the current unpriced externality regime of leaving derelict satellites in geocentric orbits eventually becomes a priced internal cost to the mission.  Reduced mass will be very important if this eventuality occurs).

The shrinkage in electronics and the entirely new server architectures made possible by CMOS semiconductor technology ate the lunch of the big mainframe and mini/microcomputer vendors like IBM, Sun Microsystems, et al.  It will happen in satellites too.

The availability of cheaper/faster/less massive technologies will allow substantial technology upgrades from the old, out-of-date, Rad-hardened silicon architectures used in previous generations of satellites and off-Earth probes.  While we don't know the lifetimes of these cheaper/better/faster-innovation technologies as applied to satellites, it is quite likely that a trade space is created to determine the economics of staying with 15-year nominal commsat life, or whether 7 or 5, or even 3, year turns might later on be the more economic option since the faster turns also allows quicker upgrades in available technologies. 

  The electronics are a small part of the mass of a spacecraft, so your premise is wrong

But most of the volume. Fiberglass has very low density. That makes the requirement for the structure which is most of the mass. Electronics get smaller and System on a Chip (SoC) designs start reducing the volume.

Edit: Or actually, more likely is that satellites stay fixed size and get more capable because the marginal cost of X more fuel isn't much once you have the launch of an ELV. We need some cheap microsat/nanosat launchers.

[Jim]

But most of the volume. Fiberglass has very low density. That makes the requirement for the structure which is most of the mass. Electronics get smaller and System on a Chip (SoC) designs start reducing the volume.

Edit: Or actually, more likely is that satellites stay fixed size and get more capable because the marginal cost of X more fuel isn't much once you have the launch of an ELV. We need some cheap microsat/nanosat launchers.

again, it isn't the electronics.  Microsats and nanosats can't be comsats.  The comm package where the mass is.   The difference between a laptop and smartphone in terms of electronics would have little effect on the mass or size of a comsat.

[Llian Rhydderch]

But most of the volume. Fiberglass has very low density. That makes the requirement for the structure which is most of the mass. Electronics get smaller and System on a Chip (SoC) designs start reducing the volume.

Edit: Or actually, more likely is that satellites stay fixed size and get more capable because the marginal cost of X more fuel isn't much once you have the launch of an ELV. We need some cheap microsat/nanosat launchers.

again, it isn't the electronics.  Microsats and nanosats can't be comsats.  The comm package where the mass is.   The difference between a laptop and smartphone in terms of electronics would have little effect on the mass or size of a comsat.

Well then Jim, looks like we should all just agree to disagree on this point.  We'll come back in ten years and see who's argument was closer to the reality we see then.   

[meekGee]

The resurgence of LEO constellation will be driven by the demand for high bandwidth, low latency, bi-directional traffic.  Kids don't want to watch TV, they want to watch youTube.

Required mass won't drop by much.  First, the reduction in mass-per-bandwidth is offset by the increase in demand for bandwidth.  Second, LEO comsats don't require less power because they are closer in.  However, they do get only 50% sun, so need twice as large panels + 45 minutes high-cycle count power storage.  They also need to worry more about thermal control, since they go in and out of sunlight.  On the upside, they don't need large fancy antennas since they footprints are naturally smaller, and don't have to be shaped funny.  (unless you consider "circular" to be funny)

So what I see is an actual increase in total required mass, but used by very large numbers of small sats instead of a few large sats.  I don't think cube-sats are the right size (even if the first constellation will use them), but maybe 100-500 kg / sat or so.  It is still after all managing a rather large cell.

At a cost level of $5M, a system that requires 1 launch/day, or $2B/yr, is pretty reasonable.  How much does Verizon spend on cell towers?

[Roga]

I did a fair amount of research on this in grad school 5 years ago, it's still relevant I think. This is known as the elasticity problem, or the valley of death. It is related to but not the same as the "chicken or egg" problem, in that it increases risk to return on investment for large capital outlays that are generally needed to develop reusable launchers.

Several people above said it was better for the reusable operator regardless because they could undercut the market and gain market share. This is not strictly true - in practice the space launch business appears to have negative demand elasticity. Demand elasticity (E_d) means, when I drop the price of a product, how much more revenue do I take in? I hope I get this right, but it goes something like this:
E_d > 1 means as I drop prices the total available revenue rises at a a higher rate than prices drop. It's difficult to keep an industry down when it has E_d>1, investment will rush to it because every incremental cost cut on the production side means windfall profits from not only existing customers migrating, but also new customers.
0>E_d>1 means that as prices drop, revenue increases but more slowly than the drop in prices. This is the normal place that most industries operate most of the time.
E_d<0 means that as prices drop, revenues also drop. Space launch E_d historically is near or slightly below 0. What this means is that if a company corners the market, it will actually make less money (revenue, not profit) by reducing costs. This is a market failure. Thanks mostly to government priorities that are even less elastic, several nations have maintained viable launch industries, but the market solves for high price, low volume, for customers who are insensitive to large price changes. Hence 3% of the total system cost.

Futron did a great survey last decade about potential emerging markets, and found very little change in this bleak picture down to something like $1000/kg. Other studies have found similar stories. Studies that focus on mass-markets tend to have much larger error bars but they seem to suggest that things like satellite servicing, personal spaceflight, space industry & R&D, and space solar power would bend the elasticity curve well above E_d = 1, but only at launch costs in the $10s or $100s/kg. Hence, "valley of death" between current $5-20k/kg and this somewhat magical point.

Additionally, there is a phase lag of 5-10 years between a capability appearing and payloads appearing. This is effectively forever at the costs of capital we see in the launch industry. At 15% discount per year, your return is worth 44 cents on the dollar in 5 years and 20 cents in 10. The current approach by SpaceX is interesting because it leverages a revenue product, so it erases most of the discount hit; but it's still money that could be spent on e.g. launching sooner. At 15% per year and 3% of total cost, slipping a launch by 3 months means you might as well give it away for free. Or more reliably.

So which way out? There seem to be a couple possibilities. First one is to make your customer more price-sensitive. Orbital and SpaceX are focusing on these customers, Orbital as a revenue generator and SpaceX as essentially an R&D investment source. Other ways might be to fly more price-sensitive payloads - humans, preferable spending their own money, come to mind. But also encouraging more standardized satellite busses where subsequent serial numbers are more and more sensitive to launch price. This is ie. Orbcomm and GPS III.

The second one is essentially self-sacrificial. While E_d is low, it's counterpart E_s is quite favorable. Which means, if you glut the market with capacity price pressure is very high. Generally that kind of thing is used to monopolize market share and raise barriers to entry. It could conceivably be used by a government or also a private player to incentivize cheap payloads. If your cost to launch is $500/kg, and your competitor is $2000, you could price at $1999 and pocket the difference. Or, you could "spend" the margin on incentives - for example, make very low base launch rates and charge through the nose for custom interfaces. Or offer significant price savings for risky R&D flights, or on reused vehicles. Or charge less for payloads that favor long-term market expansion, like on-orbit infrastructure and university or startup microsats. Or launch every 3rd day at noon with or without a payload to establish a more commoditized business model for clients and build up reliability in your vehicle.

It's a tough problem but not unsolvable.

[Jim]

  So what I see is an actual increase in total required mass, but used by very large numbers of small sats instead of a few large sats.  I don't think cube-sats are the right size (even if the first constellation will use them), but maybe 100-500 kg / sat or so.  It is still after all managing a rather large cell.

Much bigger, and fewer spacecraft, less than 100 of them

[Jim]

Well then Jim, looks like we should all just agree to disagree on this point.  We'll come back in ten years and see who's argument was closer to the reality we see then.   

All of Juno's spacecraft avionics are in this vault. A reduction in the size on the cards inside it would not have an appreciable affect on the spacecraft size.   The mass of a GSO comsat is the payload package (TWTA, receivers, solar arrays and antennas) and the rest of the spacecraft is size to this and not the avionics.

The effect (if any) is that is cheaper to launch more and not smaller. 

[watermod]

One of the great mistakes in Iridium was to use movable antennas to talk to the neighboring Iridium sats.   The movements changed balance and orbits of the satellites which caused the use of lots of extra fuel to maintain the proper orbit.  If they had used phased array antennas to do electronic beam forming instead of moving a physical antenna they would not have used as much fuel to keep their orbit.  Over the life of an LEO Iridium sat it could have been a big deal on the mass end of things.   So, yeah, modern electronics can make a big difference in the required size.

[Jim]

One of the great mistakes in Iridium was to use movable antennas to talk to the neighboring Iridium sats.   The movements changed balance and orbits of the satellites which caused the use of lots of extra fuel to maintain the proper orbit.  If they had used phased array antennas to do electronic beam forming instead of moving a physical antenna they would not have used as much fuel to keep their orbit.  Over the life of an LEO Iridium sat it could have been a big deal on the mass end of things.   So, yeah, modern electronics can make a big difference in the required size.

1.  How does spacecraft mass properties affect the orbit?
2.  Phased array antennas would require more power and hence larger solar arrays

[meekGee]

  So what I see is an actual increase in total required mass, but used by very large numbers of small sats instead of a few large sats.  I don't think cube-sats are the right size (even if the first constellation will use them), but maybe 100-500 kg / sat or so.  It is still after all managing a rather large cell.

Much bigger, and fewer spacecraft, less than 100 of them

You won't get enough coverage with <100.