Part of the reason for this blog's existence is to explore how we can get from where we are now to a permanent presence in space. I will explore that theme over a series of posts, with a focus on ISRU.
This entry covers cis-lunar space. The topic of lunar mining and fuel supply has a rich field of information available and I cannot claim to know all of it, but hopefully this will show how we can begin to harvest most of our propellant instead of shipping it from Earth.
More after the jump. First section is background information and basic design, second section is a worked example.
The tl;dr takeaway is that we could be pumping out 289 tons of propellant to LEO every year for a system startup cost as low as $2.7 billion.
First let's establish some basics:
I assume that we will develop zero boiloff cryogenic storage, reliable cryogenic fluid transfer and reliable cryogenic engine restart. Vehicles will be designed to last 20-30 years, but generally are planned to be replaced every 10 years. Obsolete systems will keep operating until they fail, providing some bonus production capacity.
The Moon's surface gravity is 1.62m/s², about 16.5% of Earth gravity.
It takes about 1.87km/s of dV to land or take off from the equator. Only a little more is needed for the poles.
Another 0.64km/s will take you from low Lunar orbit to EML1.
Low orbits around the Moon are not stable so you should not park anything important there.
Large amounts of water ice are available at the south pole and most likely at the north pole as well.
Plenty of metal oxides are available (including iron, aluminum and titanium), but carbon is rare.
In some locations the mantle interface material (KREEP) is accessible at the surface; this rock is rich in incompatible elements like phosphorus, potassium, rare earths and radioactives.
Low Earth orbit takes anywhere from 9.4 to 10 km/s of dV to reach.
Each launch site has a specific, most-efficient inclination. Changing inclinations is very expensive.
From any low-Earth orbit, EML1 is 3.77km/s away.
From Kennedy Space Center LEO, geosynchronous orbit is 4.33km/s away. It's only 3.9km/s from an equatorial orbit.
Earth escape is 3.22km/s away.
EML1 is the balance point between Earth and the Moon, the place where the gravity of each body cancels out.
From EML1, a craft with a heatshield can get to any low Earth orbit for 0.77km/s.
Without a heatshield, 3.77km/s is required.
A similar maneuver can use the Earth as a slingshot, departing from EML1 with a little nudge into any inclination and then burning at closest approach for best use of the Oberth effect.
From EML1, geosynchronous orbit is only 1.38km/s away.
Putting all of that together:
- The best place for a fuel depot is at EML1. Fuel can be harvested at the harvester's pace, shipped to the depot at the tanker's pace and accumulated for later use. Fuel is shipped to LEO only as needed and into the correct inclination.
- The surface to EML1 tanker needs to use chemical engines to overcome the Moon's gravity. It must carry about 4.6km/s of fuel, though the second half of the trip is empty and requires much less fuel. My reference tanker is 5.5 tons, cryogenic with zero boiloff and can deliver 29 tons of propellant from the Lunar surface to EML1, then land at the harvester. Each trip burns just under 34 tons, so the harvester must produce about 2.2kg of propellant for each kg in EML1.
- The EML1 to LEO tanker can use either chemical or ion engines. The low thrust option requires a very different trajectory and a dV of about 7km/s, plus a lot of onboard power. This can be a competitive option but initially it would be simpler to use the same tanker for each leg. Because the vehicle needs to aerobrake, it requires a reusable heatshield. The same tanker design is used, but it is launched with a dual-use shroud of about 6.3 tons that is kept with the vehicle. This trip delivers 44 tons of propellant at a cost of 19 tons. So, for each kg in LEO the harvester has to produce 3.1 kg. That lines up fairly well with other sources suggesting 75 tons to be harvested for 25 tons in LEO.
The harvester is an unknown quantity. There are several concepts (including one of mine), but I will use a more generic NASA number of 10kg propellant per kg of harvester per year, with 10% spares. That means if we want to harvest 100 tons of fuel per year we need to send 10 tons of equipment to start and 1 ton of spares for each year. This was for a Mars ISRU system with rover/excavator and integrated power systems. A Lunar system would be much simpler since it would only be melting and filtering water ice then electrolyzing it, so it is possible the equipment will produce far more propellant than this estimate.
Cheap propellant in LEO is useful for missions to other planets or moons. It is also useful for supplying a LEO station with water and stationkeeping fuel. Satellite servicing tugs could base out of EML1 with a ready supply of fuel and easy access to all inclinations. If commercial satellites were modified to use a water electrolysis thruster system for stationkeeping then they could be launched empty to EML1 for less dV than a direct launch to GEO. A fill of fuel at EML1 and a 1.4km/s nudge would place the satellite in GEO with decades of stationkeeping thrust available. As a bonus, the upper stage of the launch system could be refueled and repurposed. Lastly, water is extremely useful for manned operations as radiation shielding, drinking water and a reserve source of oxygen.
An example of why the interorbital exchange would be useful is NASA's reference Mars mission. Each manned trip requires three flights (one crew and two cargo), each using around 190 tons of fuel. Just the fuel would require 9 Block 1 SLS launches or 6 of the 105-ton launches; being generous we're talking about at least $4.5 billion. This is 570 tons of propellant over 2.1 years, or about 267 tons per year at around $7.9 million per ton. A full campaign of three flights to Mars would cost $13.5 billion in fuel. With lunar ISRU we can save at least $4.7 billion as described below (potentially over $9 billion) and gain a sustainable supply of propellant in EML1/LEO with minimal Earth mass.
An ISRU plant with this capacity would mass at least 83 tons and consume about 18 tons of spares per Mars opportunity. A modified version of my tanker could land perhaps 17 tons of cargo on the Moon for about 62 tons of fuel; this would require two Falcon Heavy flights (~$300 million) or an SLS block 1 flight with two additional Falcon 9 refueling flights (~$850 million). This initial 15-ton plant (with 1 year of spares and a 500-kg dextrous robot for remote repair operations) would provide the first ISRU fuel as a proof of concept. The cargo lander would be refueled and returned to LEO two months later with plenty of lunar samples. (3 tons of samples, 25 tons of fuel.) At this point the whole architecture can be validated and the final ISRU design can be settled. Until this step happens we have to use some pessimistic numbers (as below), but there is the possibility that the pilot plant will be many times more productive than expected as it will be running a much simpler refining process.
A standard tanker with dual-use shroud (11.8t empty) would be launched with a Falcon 9. Another standard tanker with 43.5 tons of fuel would be launched on a Falcon Heavy, transfer 27.8 tons of fuel to the shrouded tanker and then head to the Moon and land by the ISRU plant. The shrouded tanker would travel to EML1. At this point fuel can be transferred all the way to LEO; without a proper depot there are some inefficiencies, so only about 40 tons are delivered in each shrouded tanker (requiring two standard tanker flights to fuel up). Fortunately this is enough for the outbound trip of a cargo tug.
The first shipment of 40 tons would take the rest of the year to produce and would be delivered to the cargo tug in LEO, allowing it to deliver a 22-ton payload to the Lunar surface. (Payload would be a second 15-ton ISRU plant, two years of spares for both plants and 1 ton of other cargo.) This could be a Falcon Heavy carrying two payloads or one of several competing options. After 30 days of surface operation the cargo tug returns to LEO (again, 3 tons of samples and 25 tons of fuel).
The next 40 tons of fuel take a bit over five months to produce and allow the cargo tug to bring the next 22-ton package. Return fuel for the tug would take three weeks.
At this point the infrastructure on the moon is producing 1.25 tons of fuel per day. Each cargo trip takes less and less time to refuel (3.4 months, then 2.6 months, then 2 months). A proper depot at EML1 and three more ISRU packages round out the system. The depot is 300-ton capacity, about 20 tons of hardware and delivered to EML1 by a Falcon Heavy. In total the buildup phase takes two years and two months, almost exactly one synodic period. 90 tons of ISRU hardware is on the surface with an annual production of 900 tons and spares requirements of 9 tons. This system can deliver 289 tons of fuel to LEO every year, or 617 tons per Mars synodic period. The target of 570 tons is met with an extra 47 tons of fuel for delivering spares.
A total of seven Falcon Heavy flights are required, plus one Falcon 9. Let's call this $1.1 billion in launch costs over 26 months. Let's also assume that spares are supplied as 11-ton packages by Falcon 9, two per synodic period. A third assumption is that we will need to replace about 20% of the system every period (at a conservative 10-year lifespan for components), so we budget a Falcon Heavy flight for $150 million. That's an operational cost of about $250 million per period. In the first synodic period the fuel in LEO will cost $2.2 million per ton in launch costs. Each additional period will cost only $405,200 per ton in launch costs.
The cost of all that hardware is difficult to estimate. Let's look at two very different systems and get a ballpark figure. The BA-330 expandable space station module from Bigelow is estimated to mass about 22 tons and cost about $250 million, or about $12 million per ton. By contrast, the Iridium NEXT constellation of 72 satellites is expected to mass 57.6 tons in total and cost $2.4 billion (not counting launch costs), or about $42 million per ton. The comsats include costs like bandwidth leases and operations, but let's use that figure anyway as a first approximation. The startup phase requires 131 tons of hardware, or about $5.5 billion. Extended operation will replace about 10% of that hardware per year at a cost of about $550 million. This ignores the significant commonalities between systems; all the propulsion units are identical, all tankers identical other than the cargo adapter and all six 15-ton ISRU plants are identical.
This gives us the second part of fuel costs: capital expense. Each ton of fuel in LEO in the first period costs $9.8 million, with additional periods costing $890,000. These numbers assume we do not spread the capital costs over five periods, but front-load the entire bill into the first launch season. The first Mars caravan would cost $6 billion instead of $4.5 billion in fuel, but the next one would pay only $1.4 billion. A campaign of three flights to Mars would save about $4.7 billion in fuel costs. As a budget line item, the cis-lunar fuel network would cost NASA or a private operator $375 million per year once established.
What if, instead, a cost-effective hardware program is used that is closer to the BA-330 in terms of cost per kg? Now our 131 tons of hardware only costs about $1.6 billion with annual replacement costs of $160 million (or per-period costs of $342 million). Under this assumption, the startup phase would cost $2.7 billion with operational costs of $592 million per period ($278 million per year). Propellant would cost $5.34 million per ton in the first period and $960,000 per ton thereafter. NASA would save about nine billion dollars on three Mars flights, and would even save about $1.2 billion on the first flight.