Why are space probes so unreliable?

Now it’s starting to appear that the Beagle 2 Mars lander is another bust. You’d think that before they spend years and millions of dollars, someone would test these probes to see if they will actually respond to remote commands. Yes, I know space is a high-radiation, extreme temperature environment. But this is ridiculous. We’ve been building and launching probes for decades now, shouldn’t we be beyond the stage where single glitches scrap entire missions?

In this case it could have been not just a glitch, but maybe the probe smashed against a boulder or something; there´s a landing area for the mission, but that area is quite large and there are chances that the probe may end up on a ditch or hit a large rock; and that´s if it makes it through the reentry wich needs to be very precise or else the probe burns.
Of course it could be a design or manufacturing problem, something that made it through the test over here but gave up on destiny.

And for the glitches ruining entire missions, the problem is that, usually, such robes don´t have redundant systems, if one fails it´s over; redundant systems add weight thus increasing the costs of launch and the amount of propellant needed to get to it´s destiny; so it´s a mather of economy, or rather a bit of a gamble, you corss your fingers and hope everything stays working.

The Brits and the ESA aren’t we. This was the UK’s first attempt at a Mars lander.

Lumpy, why don’t you do it for us since you seem to have all the answers?

The answer is, that all of these things are incredibly complex devices. Mistakes like failing to have common units? Flame away. But considering that this probe was significantly cheaper than the American version that landed in 1996, it reduces the opportunity that you have to check and recheck and validate. Ultimately, you could spend your entire budget testing something, but it is the unexpected aspects which will catch you off guard anyway and you will have less money to spend on an actual product that does something useful. Ultimately, there is a chance that any probe will fail. We might as well write the probability of failure into the mission costs. So now we figure out what failed on this one and send a new one. This isn’t manned flight. Just a cost of doing business.

Better luck next time.

And it’s still a long way to go before you can assume Beagle 2 is a bust. There are many more attempts to communicate before they give up, the original plan didn’t have first attempt for 10 days.

Telemark is right. In order to hear the Beagle’s low output the antenna would have to have been pointed perfectly upon rollout.

Aside from that, It’s tough to simulate all the abuse of a ship that is subjected to a rocket launch and landing. Just a low altitude flight in the simplest of aircraft can subject the machine to a wide variety of abuses. What makes flying an airplane a safe event is the ability to correct for anomalies. A computer driven ship is limited by it’s programmed responses to outside stimuli. The parachute/airbag system sounds great until you hit a pointy rock.

You are also operating a ship with the absolute minimum of structure. Weight is EVERYTHING.

Bob Park of the American Physical Society put this in a nice perspective in his weekly “What’s New” column today (which will presumably appear at this link once his web lackeys get on the ball…At this time, it is only available in an e-mail that I alas have only at work). Basically, he pointed out that the high failure rate for unmanned missions shows:

(1) How difficult it will be to accomplish such things with manned missions where things become incredibly more complex and expensive, especially given that people are less willing to throw some money down the drain than some lives down the drain.

(2) That despite the fairly high failure rate (apparently like 1 in 3) on these sorts of missions, the amount and importance of the science getting accomplished is still much more than what is getting accomplished on each shuttle flight and at a smaller expense.

They are tested more rigorously than you can imagine. Testing usually takes more time than construction. But you can’t test them under actual flight conditions. It’s like testing an airplane in a wind tunnel and expecting it to fly across the Atlantic on its first actual flight.

We are surrounded by machines that work reliably, so we tend to forget how difficult it is to achieve reliability. The usual way is by testing them rigorously under actual usage conditions, and feeding back that information to improve the design and manufacturing process. If you’ve ever tested a factory prototype or a very specialized laboratory equipment, they are far less reliable than space probes. Unfortunately, with a space probe one iteration of this build/test/learn cycle takes hundreds of millions of dollars, and every failure gets reported on the news. I don’t see how we can avoid it though.

By the way, “we” may have been sending probes for decades but this is the first Mars probe by the European Space Agency (ESA). I believe this is also the first time the ESA attempted to land a spacecraft anywhere. At least they got the Mars Express in orbit around Mars, which is more than the Japanese managed to do on their first (and so far only) try. Give them some credit for that.

The ESA send the Huygens probe along the Cassini to land on Saturn´s moon Europa.

http://saturn.jpl.nasa.gov/overview/index.cfm

Just the other day, I finally got around to pulling my VCR out of the freezer. Thought I’d watch The Two Towers. So I stuck the VCR in the microwave to warm it up. Tripped over the cat, and the microwave, VCR and all, made a gravity assisted landing on the kitchen floor. Now the VCR doesn’t work. :frowning:
You’d think after 20 years in production, they’d have figured out how to build these things tough enough to withstand normal environmental stresses. Or maybe not. :dubious:

That’s not the problem. The problem is that it’s difficult to build such a complex machine and have it work correctly the first time without any on-site repairs or modifications. There’s only so much redundancy you can build into a system[1].

[1]Saw a good example the other day: How do you make a fuse redundant? Do you put them in parallel, so if one by accident you don’t lose power? Or in series, so you cut power even if one fails to blow? The usual answer seems to be to use 4 of them: two strands in parallel, each consisting of two fuses in series, but that’s still not perfect.

Many probes using the same sort of remote response have been tested on Mars already. This one hasn’t provided the miracle of early communications so far. Them’s the breaks.

As for the landing system, that was tested with the Pathfinder mission in 1996.

What Ale said. We do not have precise Maps of the Martian surface. We go by what Mars Global Surveyor and other orbiters have given us, and pick what looks like even terrain. Doesn’t mean there’s not something nasty waiting to trip us up.

Using the US’s landing technology, and 40 years of American and Russian results in space exploration. The people who develop these missions work together all the time. It’s perfectly reasonable to refer to the people making the collective attempts as “we”.

scr4, are you also in the habit of microwaving your VCR?

Remember, the probe was built in an extremely short time frame with a low budget. The mass spectrometer on board it was built in an engineer’s bedroom.

That’s not to say that Beagle is bust, however. It may simply have landed off course, and we are simply looking in the wrong place for it. Jodrell Bank, a huge radio telescope in Chesire, England is scanning the surface of Mars every night for possible signs. According to Pillinger, the lead scientist on Beagle, they have around 3 weeks to find it, possibly longer.

What I meant is that if you have the opportunity to test the probes under actual flight conditions over and over until you finilize the design, it’s very possible to make a 99% reliable probe. But when each test flight takes 5 years and $100 million you have to rely on ground tests, which means trying to simulate actual flight conditions. If you fail to predict the conditions, or if you neglect to test for every possible condition, you miss a problem.

The conditions aren’t that severe. We have mechanical components that work reliably in cryostats and turbojet engines, which are way more extreme than anything encountered by a space probe.

You know…with everything the probes have to go through to reach their intended targets, radiation, micrometeorites and such. I’m more surprised when one works than when it doesn’t.

Yep, as a low production volume item, it is impossible to subject space probes to anything approaching the sample of environmental stressors against which common household appliances are tested.

I don’t understand what you’re trying to say. Can you clarify?

Part of the problem is that until many probes are successfully sent, little is known about the operating environment. Thus, the first few probes are designed based on sketchy information.

Also, the strength/weight issue is an extreme constraint for spacecraft that leave Earth orbit. The faster the necessary velocity for the vehicle, the more critical is every gram of extra mass. Somewhere somebody has probably drawn a plot of structural failure rate vs. operating speed. I would guess that structural failures increase as some high power of operating speed. Airplanes have a few more structural failures than cars, orbital vehicles have many more structural failures than airplanes (reliability is about 90%), and interplanetary probes have many more failures than that (reliability is about 30%, although how much of this is due to structural failures is not known precisely.)

I think that we’re agreeing scr4. Earthbound devices are subject to failure when conditions move slightly outside the, well understood, environmental norms, or when a combination of stresses interact in an unforseen fashion. Even with the most perceptive engineers in the world, there is no substitute for the cycle of trial and error. Millions of hours of VCR testing have not resulted in a bulletproof machine. It’s far to much to expect that a few thousand satellite launches would result in reliable spacecraft.