Ever read "Colonies in Space" by T.A. Heppenheimer??? Way back in the 70s when i was in high school they were talking about a 10,000 ton nuclear power plant for the mass driver on the Moon!
Given today's political climate we'd be lucky to get some RTGs up there or even the new small reactor NASA has developed. See: Demonstration Proves Nuclear Fission Can Provide Exploration Power
Most serious mass-driver designs use superconducting coils to achieve reasonable energetic efficiency (often 50% to 90+%, depending on design). Equipment may include a superconducting bucket or aluminum coil as the payload. The coils of a mass driver can induce eddy currents in a payload's aluminum coil, and then act on the resulting magnetic field.
1. A lunar mass driver capable of launching 40 kg loads of regolith every few seconds would have to be solar powered.
2. Since it has to be based on the equator we will need remote solar power plants and high voltage cables for constant power throughout the lunar month...unless we shut down at night because of cold temps and metal embrittlement....but with enough energy we could keep things warm with heating coils. By day the mass driver will be covered by a sheet metal "tent" to prevent overheating by the Sun. Extreme temperature induced expansion and contraction of the machine could cause it to warp and deform from specifications which will be down to 0.001s of an inch. Constant temperature for the mass driver will be attempted.
3. We will have to make large numbers of solar panels on the Moon and vehicles to take crews to remote power plant construction sites. We will need long distance power cables and support poles....make support poles out of cast basalt. Cables could be aluminum clad calcium or superconductors.
4. Can we bootstrap up a base with power plants and mass driver with say a "seed" of 1000 tons mass??? Since the mass driver will use superconducting coils we must wonder....should we ship niobium to the Moon and combine it with lunar titanium for liquid helium temp NbTi superconductors? Or ship boron to the Moon and combine it with lunar magnesium for magnesium diboride 20K superconductors? Will we have the industrial capacity to do this superconducting wire making??? Or will we ship YBCO wire to the Moon? No significant qtys. of yttrium, barium or copper on the Moon but there's plenty of oxygen!!! IF we make superconducting wire on the Moon we might connect to the remote solar power plants with it...It is possible that the wire will not be so heavy that rocketing it to the Moon cannot be affordable. .but what about solid state switches? Could we make those on the Moon??? Coil supports and other massive, unitary and simple things could be made of cast basalt and non-magnetic lunar titanium. Magnet cores could be made of lunar aluminum. These will will comprise the heaviest parts of the mass driver. Cryo-coolers, Sun shields, heaters, and other equipment would be made on the Moon from in situ resources.
5. Buckets can be made with aluminum coils. The mass driver magnetic coils set up eddy currents in the aluminum and it becomes magnetic. We might as well set up aluminum production on the Moon and shoot aluminum coils away with raw regolith...then we don't have to decelerate the buckets and reuse them, something which would at least double the length of the mass driver and require a return track......the refined aluminum will find plenty of use in space construction or even be burned as rocket fuel, so shooting the aluminum bucket coils away is OKay. The bucket would probably just be glass or basalt fiber or glass fiber reinforced glass matrix composites which will find uses in space....so this will not be inefficient.
6. In space it should be easy to separate the masses of aluminum coils, regolith and buckets caught by mass catchers or will we just take all that junk and melt it down in big solar furnaces in space then run it through something like Dr. Schubert's Lunar Dust Roaster and All Isotope Separators???? With a centrifugal device we could draw off powdery regolith and with oscillating magnetic fields set up eddy currents in the Al coils and yank them away from the glass or basalt composite bucket materials, perhaps after crushing them up some.....so this part doesn't worry me.
7. We'd need some fancy machines to crank out coils and buckets every few seconds and machines to load regolith into the buckets every few seconds....at least the mass driver itself, other than the buckets, won't have moving parts!!!!
8. Perhaps we should dispense with the buckets entirely and just press screened and sieved regolith into iron molds and heat it up enough to sinter it into a solid mass....we don't need to melt it down and pour it. Then smart machines could wind aluminum coils around the solid regolith cylinder and put it in the mass driver and let it go....no bucket making, no "snap out" and bucket deceleration coils, no return track and no woven glass fiber bags of regolith. Separation of Al coils and regolith in space is simple. Just crush the stuff up and use centrifugal system to screen off the regolith then send it to the processors and manufacturing/construction space stations.
This is also of interest:
C. Site Location Although, the south pole has been identified as most promising for lunar settlement, a surface location that affords the most energy-efficient, direct injection to L2 is suggested for this example. Specifically, the equatorial site near 33.10 E longitude (Fig. 4)  has been proposed, with direct injections to L2 via "achromatic" trajectories possible from optimum sites on the lunar surface . L2 is synchronized with the lunar surface, so the injection trajectory would be fixed and therefore good for LEML. In addition, this area is also selenographically conducive to an LEML launch site, and is considered a potential site for harvesting resources for fuel production, including oxygen, glass, and aluminum . Given this launch point, the velocity required to reach L2 is 2.53 km/sec .
If a lunar mass driver launches a 40 kg payload every 2 seconds that's 1800 per hour...X 8760 hrs./yr = 630,000 metric tons per year....enough to build 4 to 40 SSPS depending on their mass....how do we feed that mass driver? What about maintenance? If it takes 10 seconds to wind a coil around a slug of sintered regolith we could make only 360 per hour one at a time....but if we have 5 machines making 5 at a time we can make 5 x 360 = 1800 per hour.....But how long will it take to load molds and sinter regolith? If that takes 10 minutes we can make 6 per hour one at a time but if we have 300 machines making 300 at a time then 6 X 300 = 1800 per hour..maybe we could get this done in only 5 minutes or less...As for maintenance, reliability, on schedule on demand performance and smooth operations we'd need 2 mass drivers so one can be shut down for maintenance or repairs while the other one works....It's possible that solid state power switches will fail and coils could short out and shut down the mass driver until repairs are made. The mass drivers would have to be built on the Moon along with all the solar panels needed for the massive energy demand of all the machinery. The cost of the imported superconducting wire might not be outrageous but the cost of rocketing all the other mass driver(s) parts to the Moon would be really high so we must build all the Massive, Unitary and Simple parts on the Moon and the complex, lightweight and electronic parts like the solid state power switches would be imported.