missinglink - 27/6/2007 11:45 AM
I have seen the Space Shuttle called the world's most complex machine. Surely this is incorrect? A nuclear submarine is bigger and has more moving parts, not to mention having a nuclear reactor inside... all the more so for an aircraft carrier? And what about the International Space Station?
This depends on how you define "most complex" of course. Simply by parts count, the shuttle is beaten by some modern airliners (mostly because of the enormous numbers of fasteners needed, though). By unique parts count, I'm not sure what takes the prize. By moving parts count, a nuclear aircraft carrier probably also beats the shuttle because it has lots of doors, valves, etc. But does having a lot of (mostly identical) doors really make a complex machine? In my opinion, the complexity of design of each part is also important in evaluating the complexity of the system.
[Sorry this turned out to be very long, but I am trying to give a flavor of the difficulties of design for space that may not be obvious to many people]
Every single part of the shuttle system is planned, calculated, modeled, tested, retested, prototyped, etc. in excruciating detail. Why is this necessary on the shuttle, and not, say, an aircraft carrier?
First, the shuttle has to lift itself into space, so everything has to be absolutely as light as possible, but still strong enough to complete its mission, even under worst case scenarios. Engineers use what is called a factor of safety in designing parts, which means that the part is designed to be a certain amount stronger (the factor of safety) than the maximum anticipated load in the worst case expected scenario in order to account for unknowns or flaws in construction. In the shuttle program (as in other space and some aviation systems) the minimum FOS is 1.4. So many shuttle parts will break at only 1.4 times the maximum loading they are expected to have to endure in order to make them as light as possible while maintaining a adequate safety margin. But because the margin is so small, you obviously have to have very high confidence that there aren't any unknowns left that are greater than your remaining margin, and your confidence that your parts are actually as strong as you think they are must also be very high.
So the flight environment must be studied very thoroughly to bound the unknowns, and parts must be thoroughly tested. Just as an example, it is standard practice for parts like screws to make them in carefully controlled batches. Afterwards, some percent of the batch is tested to make sure the chemical structure is correct, the strength is correct, etc. As another example, virtually all critical welds are X-rayed to make sure that they don't contain hidden flaws that would make the part weaker than its design implies. If you make the FOS higher, you increase the safety of the system (to a point) but you also make it heavier and less efficient, with the end result of being able to carry less payload to orbit. If you make it heavy enough, it won't get off the ground at all. This parallels the old saying: "Anyone can build a bridge that stands up, but it takes an engineer to build a bridge so economically that it just barely stands up."
Take, as a counterexample, a regular elevator (lift for those across the pond). You would think that the unknowns in elevator design are pretty well nailed down (there's no sudden turbulence in elevator shafts, for example), but as it turns out, the factor of safety used in elevator design is usually about 10. This is partly because elevators need to be safe no matter how stupid their operators are--people frequently push heavy carts into elevators or pack them with way more than the rated number of people, so this is covered by the factor of safety. On the other hand, the shuttle is always operated by highly trained ground crew and astronauts. But the other reason is because excess strength in elevator parts really doesn't hurt anything. The dry mass of the elevator is balanced by its counterweight, so there isn’t much of a performance penalty for hauling around a little extra metal. And, by being able to use standard parts with little design (for instance, using standard steel extruded I-beams for structural beams under the floor) you save a great deal in design costs and fabrication costs. In an elevator, all the floor beams are likely the same standard size (one that is sufficiently strong for all locations) whereas if you build a similar "space grade" elevator, each of the beams would likely be different. They would be just barely stronger than they needed to be and probably machined out individually from large blanks of hi-performance aluminum alloys. They would be designed in a way that they are strong enough at each location to give your 1.4 FOS for the loads at that particular location. Of course I am generalizing here, but hopefully you get the point.
As another example, take the aircraft carrier. If you want to put in a wall in an aircraft carrier, you simply have the shipbuilders weld in some quarter inch steel plate in the appropriate location. End of story. No need for computer models of the structural strength, no need for wind tunnel tests, no need to make some prototypes to test them to failure, etc. It's way stronger than it needs to be, but it doesn't matter, because the boat only has to float, which isn’t that difficult to do. There are certainly parts of a carrier that are as thoroughly studied as shuttle parts (especially in the reactors) but many aren't. The thing that makes the space shuttle so complex is that nearly ALL its parts are subject to this kind of design and scrutiny. Every single part matters in the end. In a carrier, if you want a chair for the pilot, you just add a (probably off the shelf) chair. In the space shuttle, it has to be carefully designed, tested, and built from scratch to prescribed standards.
The second thing that distinguishes the shuttle is the consequences of part failure. Because the operating conditions are so unforgiving (enormous energies expended on liftoff along with high forces, high temperatures and forces on reentry, not to mention the thermal issues and vacuum in space, etc.) failure of many parts or systems would be catastrophic. If one of the propeller drive motors on a carrier fails, it certainly isn't a good thing, especially during combat, but it isn't the end of the world either--it still floats. Either the mechanics on board can deal with it or they can airlift in parts or bring the carrier back to port for repairs. If a main engine on the shuttle fails, the consequences are much more serious. In some regimes of flight, the vehicle and crew will not survive, and in others, a risky abort is required. But even simple things can have large consequences. On a carrier, if the pilot's chair breaks, it might be annoying, but he can still stand there and get his job done, or even replace it with a folding chair from the mess hall. If the commander's chair breaks during shuttle reentry, there may be no way for him to fly the shuttle as needed. Thus, many of the parts and systems simply MUST work, and so they are multiply redundant or designed to fail-safe or fail-operational. It is instructive to take a look at a reference such as this one:
http://spaceflight.nasa.gov/shuttle/reference/shutref/verboseindex....
(or the much, much more thorough information on L2) and see how much redundancy is built into every system, and how much thought has gone into what all the possible failure modes are and how to make them redundant. There are then also systems necessary to manage all the redundancy. All this makes the shuttle more complicated.
Finally, how does the shuttle compare to the ISS? The ISS is certainly complicated, with many parts and interfaces, but in the end, it's more like airliner type complexity. The ISS is only ever exposed to the space environment--no need to design for reentry, and it doesn't have to get itself to space, which is done by the launch vehicle (mostly the shuttle), although there is much work done here to deal with the forces and vibrations of launch. As it turns out, the space environment is not TOO harsh, and pretty predictable if nothing else. Because ISS parts aren't taken up and down over and over (like parts on the shuttle), they don't need to be as weight efficient, so the design can be somewhat simpler, in some cases (obviously there are limits). It doesn't need different types of cooling systems for different stages of flight like the shuttle, it gets by with solar panels for energy so it doesn’t need cryogenic storage and energy generation, it doesn't need control surfaces for flying so it doesn’t need hydraulic systems, and on and on. And, since it doesn’t have all these systems, it doesn’t need redundancies and redundancy management for them.
Most people would admit that all this complexity is what makes the shuttle both an amazing machine, and in the end, too expansive to do what it was meant to do. However, I think it does have a pretty fair shot at being the "most complex machine."