Lunar Supply Chains?

While I often like picking fun on Mark Whittington, he does sometimes come through with some interesting links, like this one from Forbes.com: Supplying The Moon and Mars.

My undergraduate was in Manufacturing Engineering, and back in that previous life, stuff like supply chain management, manufacturing systems, lean manufacturing and all that were pretty much my forte (when I wasn't shooting my mouth off about rockets on the internet). I don't have a lot of time to comment on the articles strengths and weaknesses, but I wanted to write about it before I forgot about it completely.

Before I go any further, maybe I ought to define some terms. [Note: I know that I'm probably going to be overly simplistic here, and someone who's done supply chain work more recently than I will likely be disappointed]. Anyhow there are two broad categories of approaches for making production decisions (and a whole bunch of hybrids), sometimes simply called "push" and "pull". "Push" generally means that you are building to a forecast. You analyze what people have wanted in the past, take into consideration what you know about your system, its capacity, its availability, the dependability of your suppliers, etc, and then try to make a production plan which you then try to follow. This probably is probably still far more common than it ought to be in lots of manufacturing companies. "Pull" on the other hand uses downstream "customer" "purchases" to send information upstream about what should be made, when. Many supermarkets (theoretically) are run in a fashion similar to this. When you take an item off the shelf, and purchase it at the checkstand, information flows upstream to the next level, signaling "make one more of widget A", which requires raw materials B, C, and D, which are then "pulled" from upstream....

Anyway, if I didn't throw in enough scare quotes and qualifiers in there to give you the idea that I'm handwaiving away a whole bunch of complexity, consider yourself now more officially warned.

The general consensus these days in the manufacturing world is that the more closely you can make your company follow a "pull" production process, the better. You run into less problems with huge amounts of inventory, you end up with better customer satisfaction, less obsolescence, can run things more efficiently, etc, etc, etc. The problem is that in order to transition towards a pull production process, you have to seriously improve the availability, reliability, and capacity of your system. You have to make serious changes in how you work with your supply chain, etc. Most companies that have tried and failed to implement "pull" production systems (whether they call them JIT, Lean, or whatever the buzzword of the month is), have failed to do so because they tried to go directly to a "pull" system without first understanding their current state and finding out what things need to be fixed first.

You know, I could go way more into detail on this, but I can already hear people snoring and I haven't even posted this article yet. So I'll get back to space stuff and gloss over a bunch of the Manufacturing Engineer nerdlichkeit.

So, a couple of the points they make in the article that I thought were good:

• Prepositioning of supplies makes a lot of sense for lunar and martian missions. There's no reason why all the equipment for a lunar landing, for instance, should have to fit into the same vehicle as the astronauts coming over. Of course, if NASA really took that to heart, it might undermine some of their faith in the Cult of Heavy Lift.


• For Mars transportation, so long as flights take as long as they do, it's impractical to do too much in the way of "pulling". However LEO is definitely close enough (travel-time-wise) that they should be able to "pull" supplies there.

That said, I disagreed with two things:

• They state that for Lunar resupply, the times involved are enough to require a "push" from LEO. I have to disagree, The Moon is definitely close enough to allow for "pull" type processes, if done right. When you think about it, the trip time from LEO to the lunar surface, even if you go through L1, is about on par with that of a UPS ground shipment from say Connecticut to California. Admittedly, with near-term transportation systems, you can't "overnight" stuff to the moon from LEO, but it's definitely close enough for pull systems to work reasonably well.
  
  I think these guys are falling into the same trap that they criticized at the start of the article--assuming that all shipments to the moon have to be made by big vehicles launched from earth. If you assume that all of your logistics has to be shipped up by a cargo module stuck on top of an EDS, launched by a CaLV, and landed by an LSAM, then I guess I can see their point. NASA will be lucky to fly more than 3-4 times per year (if that!) with their CaLV. If you assume that's the only boat in town, then effectively you get four flights to the moon per year. It's really hard to run a "pull" system with that much of a time-lag.
  
  But I think this points out exactly where great improvements could be made. Basically, the "move batches" need to be broken up as much as possible. What I mean is, that if instead of waiting until you had enough cargo to fill a 30-40 ton lander, why not ship it up in smaller pieces on a much more frequent basis. Say shipping it in 10 ton increments, or 5, or even 2 tons. Probably 98+% of the things needed on the moon are divisible efficiently into chunks of less than 2 tons. Say the total demand for cargo stayed the same in that situation, at 160 tons of cargo per year to the lunar surface. If you borke that up into 2 ton chunks, that would mean a flight would be going from the earth to the moon every 4-5 days. If the cargo shipment interfaces were well enough defined by that point, it would mean that you could likely place an order on the moon, and have it delivered within 5-10 business days. [Note: this is one reason why I feel that ion-powered tugs really don't make much sense for all but the biggest of lunar cargoes. Sure, the cost of stuff shipped up from LEO goes down, but if you have to wait 4-9 months before something arrives, it better not be important. Not to mention you have an expensive ship and cargo that are basically sitting around tying up capital like inventory for the whole duration...] There would still likely need to be some inventories of critical spares and such, but with a transportation system like that, you could run much closer to an ideal "pull" system, and it would likely be a lot more efficient.


• The other thing I didn't like was the assumption that trips to mars can only be done every two years during the "launch window", and that they have to take 6-9 months. While that is definitely the most mass efficient way of doing things...economically speaking it's probably a very inefficient way of doing things. If Mars is to ever become an economically important part of humanity (or anywhere else off-planet for that matter), regular deliveries are critical. Having supplies show up only once every two years just will not work. The amount of emergency spares and other equipment inventories that will be needed is just mind boggling. Think about it. It could be as much as almost three years from when you discover you need a part until when it could get shipped from earth if you use their assumptions about the transportation system.
  
  Once again, I think that a lot of the limitations they're running up against are due to having to assume a crappy transportation architecture. Yes, there are physical realities involved. Traveling to Mars at a time other than the biennial launch windows is a lot more delta-V intensive. Traveling to Mars in less than a 6-9 month shot is also a lot more delta-V intensive. Those are cold, hard, physical facts. Now, I'll admit that I don't have exact numbers on how much of a delta-V hit you have to pay to launch on an off-nominal time, but I'm sure someone reading does.
  
  What I would suggest for this is similar to my previous suggestion--break things into smaller, more frequent chunks. If chemical propulsion is all that's around when this starts happening, you could go with a modular architecture. Basically, imagine having a modular set of reusable tugs, which you can vary the number of tugs compared to the cargo freighter mass and what time in the 2-year cycle it is. You can add extra parallel stages, or carry less cargo during those times when it is between launch windows, and use less stages per flight, or carry more cargo when the timing is more ideal. All (or maybe all but one) of the propulsion stages for any given cargo flight return themselves via propulsion and/or aerobraking to LEO for reuse several times within a 2-year cycle. Eventually once you have propellant supplies on Mars, you setup a similar set of propulsion tugs on that side. You can make up for the less mass per individual flight during the "off season" by flying more frequently. Sure it costs more that way, but if you can offer more regular flights, that will more than likely pay for itself in less require inventory and stockpiling there on Mars (not to mention keeping as much of your transportation capital "turning around" on a monthly or bimontly basis instead of biennial also sounds pretty nice too, no?). I'm not sure about this, but if ion propulsion can get something to Mars in similar amounts of time (leaving from say L1) as chemical propulsion, then maybe that might make a more useful market for it than cislunar transportation. If your trip already has to take 4-9 months anyway, why not be more efficient and use a propulsion system that might be less sensitive to "launch windows" anyhow? I'm not sure, about that though. Maybe Dennis could comment.

Anyway, my point is just that a lot of these supply chain modeling tools are only useful if you remember that that they are most effictive if you use them to figure out what assumptions need to be changed, and how. If you assume that your assumptions are fixed and unchangeable, such modeling tools will allow you to make more out of a lousy deal, but their true potential comes from helping you figure out which assumptions need to be questioned, challenged, then ultimately changed.