6 thoughts on “Bigger Is Not Necessarily Better”

  1. In the table in Horowitz’s article, how can Ares V, with six liquid engines and two solid engines on its first-stage, possibly have a lower calculated per flight failure rate than an EELV Heavy, with only three liquid engines? (0.99 for Ares V versus 0.97 for EELV.) Just looking at the number and types of engines alone (eight versus three total engine, two different types of engines versus one), there’s just no way that Ares V can beat EELV’s failure rate. Horowitz’s numbers and article are goofy on this basis alone.

  2. Joe,
    Also note that the reliability he quotes for the Ares-1 is nearly 25% higher than the ESAS numbers (which were before issues like thrust oscillations, and having to strip redundancy off of Orion and first stage systems came to light). And the numbers he quotes for EELV LOC numbers are nearly 10x lower than the ESAS numbers…


  3. Ok, I see how the LOM numbers were calculated for the Mars mission. Assume 15 launches are required to support a Mars mission, launches have a 97% chance of success, and the mission fails if any one of them fails. Then your chances of success for the launch phase of the mission (ignoring the risk in the rest of the mission) is a bit over 63% and the chance of failure is 1 in 2.7 as predicted.

    Interestingly, the chance of LOC depends on when the crew launches. If it is the first thing in space, then the LOC number of 1 in 133 counts. But if the people are the very last thing launched, then there’s a sizeable chance that the mission fails prior to this point and the people aren’t even launched. That brings LOC down to 1 in 200 roughly, but only because there’s a good chance you don’t even bother to launch the crew.

    Let’s assume the 97% failure rate per launch of the EELV is correct.

    Suppose we have a backup copy of everything that’s to be launched, including the crew and ample launch capability. Then for each of the 15 payloads, we also have a single unique duplicate payload (and assume all the payloads are distinct). So what is the LOM in this case? For a single pair of payloads, we now have 99.9% chance that one of the two payloads makes it to space, the chances of two rockets failing in a row (assuming that the failures are independent events) is 9 in 10,000.

    The odds of 15 successful deliveries of payload (and our chance of LOM) now go up from 1 in 2.7 to 1 in 75. If we have three of each payload (including three crews) the odds go up to a LOM of roughly 1 in 2,500. The LOC number remains stubbornly fixed at 133 though we now have a small chance of multiple losses of crew. Yes, the LOM for the mission is higher than the LOC for the mission. Of course, that’s assuming generously, that someone is bold enough to launch additional crews after the first crew is lost. For a timid mission manager who aborts the mission after LOC, then of course LOM will remain below LOC.

    Suppose we had an infinite supply of all payloads, including crews, and no scheduling difficulties no matter how long it takes us to put everything in orbit. How many launches on average would we expect to do in order to make a complete vehicle and crew in space? Turns out to be 15.5 roughly. In other words, all this extra launch redundancy only costs us half a launch on average. LOM no longer makes sense in this extreme context because everything gets into orbit eventually. Yet the LOC remains at 1 in 133.

    You should be able to do something similar for the Ares vehicles. The only problem is that additional launches of the Ares V are much harder to come by. NASA is scheduling 3 Ares V launches a year. That’s not going to give a lot of slack in case a launch fails. Still if they have even one extra Ares V launch available, they can improve the LOM for the Mars mission to 1 in 99.

    In summary, EELVs with two backup copies of each payload including crew can have far better LOM numbers than a far larger, more reliable rocket with no payload redundancy.

  4. The only way you’re not going to have that kind of redundancy in payloads is if your program is so marginal you’re only planning a few missions (or even one).

    Any solar system exploration program that would be worth doing would have the payloads rolling off production lines, not hand crafted in limited batches.

  5. Paul and Karl,
    Not to mention that if most of the flights are just shipping propellants up, you’re talking about replacement cargoes that you can order from Praxair for delivery next week, not long-lead-time engineering articles.


Comments are closed.