lmike - 21/12/2005 8:14 PMOf course, the system should be able to do an auto abort, the whole point of the electronics on the LES is the fast reaction time, especially with solids there are less things to monitor for clues. Or, if the crew are incapacitated. The manual control is there just to force an escape, not to prevent it. I'd also give this ability to the ground controllers, in case there are some visual clues from outside on launch. There could be (extremely unlikely) scenarios where the system could do a false abort in the extremely short time that the LES is active at all, 2 minutes of the ascent?, but at the worst the crew would survive, just with a loss of a good vehicle.
...The bigger problem, I think, is how to make a reliable abort decision based on a few sensor readings. But with some statistics of what can go wrong with a solid first stage most of which has already flown some 200 times, and a couple of sensors here and there it's not that hard really. Nothing that requires an AI, or a high level language, that's for sure.
Jamie Young - 21/12/2005 1:07 AMYou sound like you know what you're talking about. What do you think would be a better choice of software/lanuage, whatever this is? I know nothing about this subject.
Dobbins - 21/12/2005 9:36 PMThe use of a solid booster first stage makes the LES harder to design and program. In the Mercury and Apollo systems the LES could shut down the booster if it was out of control. A solid rocket can't be shut down so the LES has to be capable of reacting to any failure modes that produce side thrust.The Mercury LES didn't even use a computer, it's sequencer relied on relays and logic circuits. You could get by with something that simple on a liquid fueled rocket, though it was a nightmare to checkout. A LES for a solid will require a fairly complex program.
lmike - 22/12/2005 12:08 AMyou are riding a firecracker basically... hence, no time for loading instructions/dereferencing memory/handling exceptions (gosh forbid)/running a shortest path algorithm in a decision net/running a state machine/... If a bunch of 'if...then...' at the circuit gates level can do it (like say transistor relays/gates), then so it should be done*. As simple *as possible*, but no simpler. And the complex logic requirement is nowhere in sight here.*it's not like they are going to need to debug/reset the thing once in flight(unlike the MER rovers, for example)... Hopefully...
SimonShuttle - 22/12/2005 5:00 AMRealtime, you mentioned about the Shuttle's systems being very much debugged over years. Is that a bit like having an old computer, but because there's no need for constant Windows patches it can run smoother than a newer computer?The Shuttle flight computers seem to work very well, talking to the engines and back etc.etc. But some people redicule how old they are. Is that the case then, they work a lot better than their age may suggest because they are so debugged then do the job better than sticking some Window XP computers into the Orbiters?
darkenfast - 22/12/2005 1:58 PMRegarding the need for speed in aborts, I believe the danger may not be quite as severe as some would imagine. As I understand it, neither stage of the Crew Launch Vehicle is capable of "detonating" (as opposed to a fireball like Challenger experienced). This is very well explained in the "Reliability and Crew Safety Study" done for ATK by SAIC (don't have the link handy, but it's easy to find). The other thing is that a capsule is much more likely to survive either a fireball or aerodynamic loads from a vehicle breakup than something like the Orbiter, so even if the system isn't the fastest (in computer terms), the crew will likely survive. Remember, it was the aerodynamics that actually destroyed Challenger. I do believe that it's very important to have manual actuation, in addition to the automated system.