click for video clip click for clip or download avi

For a number of years now I've watched the progress of John Carmack's experimental rocket company Armadillo Aerospace, as have many others. We've seen Carmack abandon hydrogen-peroxide rockets and move to a more conventional liquid oxygen / alcohol fuel. We've seen fantastic videos of large scale hardware hovering seemingly effortlessly in heavy winds. We've seen some spectacular crashes, too.

This year, Carmack and his team have been focusing on the Lunar Lander Challenge. At the time of writing, they are considered the hot favourite to win the event, having been the only team to be even remotely close to fielding a vehicle last year.

Unfortunately, I think maybe all this success is going to Carmack's head. The presentation for Space Access 2007 included the snippet above which attracted not only a few cocked eyebrows but some overt bursts of laughter (or so I hear).

Being a fellow computer geek I have this skewed view of how things scale that makes Carmack's presentation here look almost reasonable. The way I see it (and the way I guess Carmack sees it), he has a working design that is stable and at a certain, presumably high, quality level. He wants more thrust than a single module provides, so why not strap a few modules together?

Well, apparently, rockets just don't scale like that. I guess Carmack is of the opinion that if one of his rockets fails then the others will just have to work a little harder to take up the slack. Maybe he sees it as a sort of a redundant array of rockets.

Thing is.. when rockets fail, they don't tend to just quietly shut off.. they tend to explode.. and that tends to cause other near-by rockets, and fuel tanks, to also explode. All the rocket scientists in the audience who were chuckling to themselves as Carmack made his presentation, of course, know this, and that's why the prevailing wisdom of rocket building these days is that it is better to make one really big rocket than to strap together 5 or 6.

Well ok, fine, it's a bit against the conventional wisdom, and that means that Carmack will have to work against the grain a bit, but does that mean it is ludicrous? Perhaps the most safest rocket ever built, the rockets used to lift the russian Soyuz spacecraft, the R-7 family has 16 engines (chambers) in the first stage and 4 engines in the second stage.

But perhaps the greatest support for the feasibility of Carmack's wacky suggestion is the, still yet to fly, Falcon 9 from SpaceX, which has 9 engines in the first stage and a single engine in the second stage. Maybe this has something to do with SpaceX's founder, Elon Musk, also being a computer geek like Carmack, but if the Falcon 9 flies well in Q2 2008, maybe Carmack's modular rocket also has a chance.


On 31 May 2007 I received this reply. The sender permits me to publish it here, so long as I keep it anonymous..

I've been following Carmack's rockets for a long time, particularly on the aRocket list. I'm not an Aerospace engineer, but Carmack's a really, really smart guy. He may end up being wrong with this crazy idea, but I don't think so. Some of the points you raised:

Thing is.. when rockets fail, they don't tend to just quietly shut off.. they tend to explode.. and that tends to cause other near-by rockets, and fuel tanks, to also explode. All the rocket scientists in the audience who were chuckling to themselves as Carmack made his presentation, of course, know this, and that's why the prevailing wisdom of rocket building these days is that it is better to make one really big rocket than to strap together 5 or 6.

First, "rocket scientists"... while one should have respect for their expertise, they are also the ones who are typically first in line to say that things "can't be done" cheaply, mostly because of the way Aerospace has been funded and done the last 40-50 years. Of course, they have history on their side since that's the only way it's EVER been done...

But it's interesting to note that the insurance company that put up the prize money for the X Prize hired a "big space" engineer to advise them on whether the X Prize could be done, and they were assured that it was impossible.

Now, onto the point above. Yes, rockets can explode and might take out all the modules, but an explosion would take out a rocket whether it's modular OR singular. Of course, more rockets = higher probability of failure.

But here's where Carmack is truly a genius: he cares more about simplicity and reliability than performance. The typical Big Aerospace rocket is built with very high performance, but also with extremely low tolerance for failure. On the other hand, Carmack has made a lot of posts where he describes intentionally stunting the performance for practicality reasons.

One example is that his rockets will stay subsonic in the thick air so he doesn't have to worry about aerodynamic issues. The "real" engineers on the aRocket list thought this was incredibly wasteful, but it's actually brilliant. Fuel is cheap, and he doesn't need the extra performance.

Another example is that he's planning on his orbital vehicles to go straight up, then shoot off at an angle instead of the typical sloping trajectory. This is again wasteful of performance, but it also simplifies a lot of things, as well as simplifying regulation (he doesn't pass over anything other than the launch site on the way up). Again, some other engineers were appalled by the wastefulness.

The bottom line is that, yes, he has more rockets, but his rockets are going to be far more reliable than the usual rocket.

Think about it this way: how dangerous is a big jet engine with that spinning turbine? If it exploded, it could take out an entire airplane! Yet people sit next to them all the time. It's because they are built extremely reliably. I'm sure they could get a lot more performance out of them if they risked explosions 5% of the time, like the typical rocket.

We have a long ways to go, but the longer I've watched Carmack do his thing, the more impressed I get. The achilles heal of Aerospace Engineers (and most engineers, for that matter) is that they try and design maximum performance without looking at the whole picture of reliability, safety, regulation, etc. Carmack is smart enough to realize that performance is a low priority in his problem set.

If you're not a subscriber to the aRocket list, I highly recommend it:

http://exrocketry.net/mailman/listinfo/arocket


25 May 2009, Update.

Well, last October Armadillo won the Level 1 top prize of $350,000. It is expected that this year they will win the Level 2 top prize of $1,000,000. To date, Armadillo have done no scaling of modules. Most of their efforts have been directed towards using their rockets in the Rocket Racing League (this has now been cancelled) and work with methane propellants for NASA. The later is concerning to me. It seems that NewSpace companies inevitably become "dino space" companies after they take government money. Carmack has also been busy with keeping Id Software alive and well with the economic downturn and all.

All that said, a recent update indicates that they will soon be doing some free flights to some significant altitude. These will presumably be single module flights, but perhaps with the Level 2 prize money in hand they will finally start doing some scaling.


If you'd like to comment on this page, or any other page on my web site, send me an email.


QuantumG
<< back to my home page