Re: Upcoming SpaceX Falcon 9 Flight 1 Launch
I agree. The NASA approach, known by the term "failure tolerance", is essentially to make every system redundant in an attempt to prevent the first mission from having any risk. This seems logical to people without real hands-on experience, and is one of the fundamental requirements for the fabled "man-rating" but it is ineffective for three reasons:
1) Redundancy increases reliability only if failures are random. This can (rarely) occur during long mission phases, like the years the rovers spend on Mars, but launch lasts only a few minutes and there is no time for components to "wear out". Despite the high risk of launch, almost all launch failures are deterministic. A classic example is Challenger. It was "failure tolerant" because it had both primary and backup O-rings. Both failed simultaneously.
2) The only failures that can actually be mitigated are those that are anticipated, and if a failure is anticipated it can usually be avoided by a feasible design change.
3) The vehicle is saddled with the mass and cost of redundant and "failure-tolerant" systems for its entire operational life.
Thee are certainly times when redundancy makes good sense. But when requirements are discussed, I sometime bring up the question of whether the system reliability has even been accurately measured. If the system is reliable, redundancy may not be needed, while if the system is unreliable, redundancy may not help. Usually I have been politely ignored. Reliability is tedious to measure and people want instant answers.
What few in NASA remember is that the "all-up" systems engineering approach, where failure analysis and mitigation is all done on paper before the first launch, was adopted only during Apollo and only because it was a "crash program" under extreme time constraints. The traditional approach in aviation is to test components and systems extensively, adopting changes incrementally and only after they have been thoroughly tested in actual flight. That way, design problems can be identified and corrected without adding tons of redundancy. In the words of A-4 designer Ed Heinemann, "Simplicate and add lightness". It would be much better for SpaceX to lose a few early launches and end up with a design that is simple, reliable, and less expensive. Of course, it wouldn't make sense to put people on the vehicle until it has flown a dozen times or so, and any significant mod should be flown unmanned at least once.
However, even as we speak, there are those who denigrate SpaceX and indeed all "commercial" launch vehicles because they don't follow the NASA strategy. Yet Soyuz gets a bye on the whole mess. As a former industrial engineer, I sometimes tear out my hair.
Facebook
Twitter
Instagram