Printable Version of Topic

Posted by: mcaplinger Nov 14 2007, 03:26 AM

http://www.msss.com/msl/mardi/news/12Nov07/index.html

Posted by: nprev Nov 15 2007, 01:20 AM

Sad news, but a good decision. I hope that integration testing in ALL areas is still aggressive & ongoing while there are still six months or so to do something about any other unexpected problems.

Posted by: mcaplinger Nov 15 2007, 02:23 AM

Out of curiosity, is there any increase of failure probability that you would have judged acceptable? E.g., what about a 1 in a million chance, or 1 in a billion?

People who say that the risk of doing something is 0 are probably just not thinking hard enough.

Posted by: nprev Nov 15 2007, 02:35 AM

Wasn't so much the probability of occurrence as the nature of the conflict. Potential interference with IMU inputs during EDL (where, obviously, they're needed the most) is scary indeed, and I'm sure that the Phoenix team is not at all anxious to repeat the MPL disaster (which, admittedly, arose from completely different root causes; the point is the same).

Losing the imagery is far better than accepting even a very small risk of losing the vehicle, so gotta vote WAY on the side of caution here; think they did the right thing. We're all well aware of how problematic landing on Mars has been historically; don't think that we need to take unnecessary chances. To quantify a little bit anyhow, here's where I'd put this on a probability vs. consequences plot, and it ain't makin' it.

EDIT: Just to clarify, I am in no way trying to denigrate MARDI or its science return. Context data is invaluable, and it's really a damn shame that we won't get it from an instrument of this quality. However, given the fact that the spacecraft is in flight and no work-around appears evident, mission assurance needs have to take precedence.

Posted by: ugordan Nov 15 2007, 08:37 AM

Except that in this business there's always at least a "very small risk of losing the vehicle". Most of the time it's much greater than 1 in a 1000, realistically speaking. Personally, I think this decision was influenced by too great of a paranoia. There's trying to lower the realistic risk beyond that what's feasible and then there's trying to lower the mathematically calculated risk. This decision seems to me to be more of the second one in nature, not that much is gained while quite a bit of nice science is lost.

Posted by: mchan Nov 15 2007, 11:57 AM

From the release:

"changing the timing within the EDL sequence would have necessitated additional testing, and arguments were raised that the human and time resources for this testing were in short supply."

Could human and time resources have been allocated to do adequate testing? What would it have taken (money, borrowing people from other projects)? Sounds like a management call.

Posted by: tedstryk Nov 15 2007, 12:04 PM

I believe the problem was discovered with only a few days until launch.

Posted by: mchan Nov 15 2007, 12:41 PM

The original problem of potential interference between reading MARDI and IMU data was found before launch. Then there was the fallback to taking one image which would have fit in the MARDI internal buffer. But the there was the 2nd problem of how to take one image with minimal risk. The evaluation of the 2nd problem led to dropping even the one image from the landing plan.

Posted by: djellison Nov 15 2007, 01:24 PM

The cut from many images to one image was nasty - and I admit I really feel sorry for the MSSS guys who've lost two MARDI's now. The cut from one image to none in some respects makes a bit of sense, if the funds can be freed up to fly MARDI as a result. One PHX-MARDI image, or an HD MSL-MARDI movie....that's not a hard call to make imho. Of course, ideally, the answer would be 'both'.

Doug

Posted by: tedstryk Nov 15 2007, 02:31 PM

This reminds me (on a much smaller scale, since with HIRISE, the science to be gained from MARDI isn't in the same ballpark as this example) of the decision to not use the remote sensing instruments during Galileo's approach to Jupiter after the tape recorder anomaly. Wes Huntress angered a lot of people with his decision. Granted, given the performance of the tape recorder for the rest of the mission, the data probably could have been taken. However, had the tape recorder had another glitch and the probe data been lost (or worse, had it snapped outright and essentially ruined the LGA-based mission plan), there would have been hell to pay, and on the heels of Mars Observer, which had failed only two years earlier, may have been catastrophic for the entire planetary program. Chances are, taking the MARDI images would be fine. But given the coverage the mission would get if they ignored a known glitch and crashed (even if it was later proven not to be the cause). I think the only way a successful Phoenix (or the MER, for that matter, since they are still kicking...um, I mean rolling) would get that kind of coverage is if they photographed a martian dog walking up and urinating on the lander's leg (or rover's wheel - just walking by probably wouldn't get as much coverage as a failure).

Posted by: stevesliva Nov 15 2007, 03:44 PM

And sometimes managers think that 9 women can have a baby in a month...

Posted by: Greg Hullender Nov 15 2007, 06:46 PM

Also, try to remember that even estimating risk is difficult. Especially for one-of-a-kind events.

--Greg

Posted by: SpaceListener Nov 16 2007, 01:26 AM

How will Phoenix land with a reasonable safety on Mars without the asist of MARDI? Perhaps, it will use the altimeter rader to measure the height from the surface?

Posted by: mchan Nov 16 2007, 02:53 AM

MARDI was never required to support descent and landing operation. You may be thinking of DIMES in the MER EDL which was intended to remove excess horizontal velocity. Phoenix uses a Doppler radar for vertical and horizontal velocity measurements which feed into the descent control loop.

Posted by: nprev Nov 19 2007, 06:04 PM

Oh, yeah, man, no doubt; should've clarified that comment a bit. What I meant was that it seems unwise to accept a known risk in addition to the unknowns and the uncontrollables. If MARDI was essential for EDL, then by all means this would be an acceptable risk; since it's not, then mitigating it is appropriate.

Not that I'm not severely bummed... ...was really looking forward to those images, and I feel bad for the MARDI investigators. Sometimes systems engineering is anything but kind...

Posted by: tuvas Nov 20 2007, 04:52 AM

You want to know something that's scary. There are 12 thrusters on Phoenix. It's public knowledge, but not well known, that if one of the 12 fails, the landing will fail (Unless maybe if it were at the very end...) Scary, huh? There's risks in everything in spaceflight, but, well, I guess sometimes something has to go...

Posted by: nprev Nov 20 2007, 05:38 PM

Well, that's an example of a completely unavoidable risk; there's just nothing that can be done to mitigate it other than make sure during the design phase that the thrusters are as reliable as possible. The sneaky, scary stuff comes up during systems integration & testing, and with planetary missions there usually isn't enough time to accomplish all the testing you'd like before the launch window (to say nothing of the budget to do it!) That's a big part of the unavoidable risk assumption as well.