Kepler Has Problems
NASA Kepler Mission Update 17 January 2013
“Earlier this month during a semi-weekly contact with the spacecraft, the team detected an increase in the amount of torque required to spin one of the three remaining reaction wheels. This increase in friction occurred before the Jan. 11, 2013 quarterly roll, and persisted after the spacecraft roll and several momentum desaturations of the reaction wheels. Increased friction over a prolonged period can lead to accumulated wear on the reaction wheel, and possible wheel failure. To minimize wheel friction, the team implemented several mitigations including increased operating temperatures, higher spin rates, and bi-directional operation following the failure of reaction wheel #2 in July 2012.”
Keith’s note: If a second reaction wheel fails, the mission is over.
Deep breath.
Why is the mission over if a second reaction wheel fails? I’m interested in the details of how it all works.
Tyler:
Kepler is a ‘small’, relatively uncomplicated telescope that must point at the same part of the sky all the time. Because of the pointing accuracy needed, Kepler probably couldn’t manage on just two moment control gyros like the venerable SOHO Sun facing satellite.
Most satellites like Kepler have 4 of these gyros. One for each direction, yaw, pitch and roll, and a spare that can be repositioned to replace any of the other three if they fail. This only accounts for one failure but, as we see with Kepler, can save a mission from complete shutdown.
As for the ‘small’, relatively uncomplicated part, Let’s not forget who built it or how it was funded. Recently, a small company displayed a prototype of an asteroid hunting satellite of about the power of Kepler that they claim they could manufacture for 5% of Kepler’s cost. They would leverage ‘off the shelf’ parts, lean, mean innovation and ‘standard business practices’ to achieve this. Much in the same way that SpaceX designed and built Falcon and Dragon.
So, even if Kepler blows another gyro, all is not lost! We will have a unique ‘planet hunting gap’ to fill though. Like human space flight, it’s not something we can do without anymore.
tinker
That helps, thanks tinker.
Except that there are a couple of major errors in justatinker’s post. He (or she) is confusing momentum or reaction wheels (attitude control systems) with gyros (attitude change detection systems). Reaction wheels turn electrical energy into rotary motion and thus torque, or vice versa (they’re like flywheels).
Second, SOHO has no operational gyros, nor has it had any since February, 1999. It senses attitudes with sun sensors and/or star trackers. Its wheels, however, are fine, and it’s still operating on the wheels it launched with over 17 years ago.
Third, three of the wheels are mounted with perpendicular axes (call them x, y, z) and the fourth, spare, or “skew” wheel spins around an axis inclined at 45 degrees to all the other three, so it can be used to replace any one of them, but it can’t be “repositioned.” Lose any two of the wheels, however, and the spacecraft can no longer cancel out all its angular momentum to remain pointed.
Want to really confuse them? just mention Control Moment Gyros (CMGs), like on fast pointing Earth observers….(e.g Worldview,2,3…).
Kepler had the 3X orthogonal wheels plus the “spare” as you mention above. Lose one more and it will not be able to point at Cygnus.
With solar pressure balance and two wheels it might still be able to do some stellar variation science on other (fleeting) targets, but fuel would run out very quickly compared to just doing desats.
Kepler is already a bit past its initial official planned lifetime of three and a half years, so a failure soon would still be considered a success. And it’s certainly produced a huge amount of science. Still, it would be too bad to lose it now. NASA’s science missions have been teaching us to expect spacecraft (and rovers!) to last far beyond their initial planned lifetimes.
And we want to see more transits for confirmations of the smaller and not as close to the star planets. It would be wonderful to get 3 more Keplers up with modest improvements and pointing at different spots.
How much would it cost to replace Kepler? Preferably using the same design.
The current one cost about $600m at launch. I’m not sure about ops budget, but extending its mission beyond 2012 was apparently estimated at $20m/yr. (That seems a bit light.)
So a new spacecraft every four years for about $170m/yr, assuming no savings from duplicating the design, nor improvements to the spacecraft’s lifespan from incremental development and flight-proving hardware. Bargain.
Which is why it won’t happen.
If there was support for a new Kepler, the higher-ups would immediately start pushing for an all-new design, bigger and better, involving the most state-of-the-art systems they can think off, with everyone throwing their wish-list of tech-development into the pool. And instead of a simple $170m/yr Kepler replacement, you’ll get a multi-billion-dollar JWST-type monster.
[Edit: See? It’s already happening in this very thread! “its time to push the bar forward with a follow on to Kepler that expands on what we learned with Kepler. Whether that is TPF or something else, lets not no “re-do” the mission.”]
If there wasn’t enough support to enable them to twist the program into a JWST-style monster, then the proposal would be killed in order to keep the actual JWST-monster fed.
Since the JWST will not have been completed there may be a sufficient number of worried people to ensure that Kepler 2 is restricted to parts with a TRL 6.
There’s a big wide gap between expanding on what we learned with Kepler and throwing every questionable gizmo onto the mission.
Yes, there’s a big gap. But look at NASA history for a while and you’ll see that over and over again they’ve done exactly as Paul describes.
Words like “every” and “everyone,” of course, are to make a point and not intended to be taken literally, but the problem is real.
The original “Mars Reference Mission” comes to mind. By the time they had included everything that all of the contributors had declared was essential, the price tag was $450 Billion — yes, BILLION — and that was decades ago now. In reality, that $450B included a lot of stuff that was in no way essential; it was just a collection of people’s pet projects that they tried to fund by tacking them onto a mega-program. The same thing happens on much smaller programs, too.
It seems to be an ongoing NASA problem that proposed programs have costs that are either too big or too small to be realistic.
The flip side, which I alluded to in my last paragraph, is that any project that doesn’t play this game won’t have enough support to avoid being cut (or never funded) to pay for cost overruns in another program that does play the game. You just can’t win.
Just mentioning continuing Kepler as a long term iterative observation platform ($170m/yr, bargain!), brought out the criticism that it’s done, it’s over, yawn, boring, build us a new shiny thing! Is it any wonder that NASA can’t develop a culture of incremental development?
Paul,
Having agreed on the symptoms of the problem, I have to wonder, in all seriousness, what would happen if a string of NASA programs were “properly” planned, costed, and then executed as per plan. Would this help to “fix” the system, or would the bottom feeders fight back harder to sustain their pork supply? It can be done right, but where are the incentives?
Steve
I think there are obviously people at NASA trying to fight back. I think the creative use of SAA’s and support fixed price commercial services comes from some people trying to sneak in a better way of doing things. And you do occasionally see small programs runs well (Kepler itself seems a reasonable example) but those programs don’t seem to be rewarded.
But it seems like human instinct itself is against sensible staged development. We want the shiny new thing, because the “106th exoplanet” is boring. And the sunk costs fallacy mean we have to “accept mistakes and move on” with JWST/SLS/Orion, we reward failure and punish success.
Big Rocket is best rocket. And Orion is best capsule because it is the most “advanced”. And JWST is “ground breaking” and therefore worth choking out dozens of other potential science missions. And the shuttle was the “most complex machine ever invented” as if that is a good thing.
It stuns me that the incremental development of early space-flight lasted as long as it did. How is it that NASA didn’t go straight from one Mercury flight to a ten year gap trying to develop Apollo in one step? Or straight to Voyager (yes, singular) without all the Pioneers, Mariners, etc?
[Joe said: “How do you incrementally go from Kepler to something that does something other than what Kepler does?”
In the case of Kepler, the improvement would be technological rather than task. Indeed, the value I give to Kepler’s continuity is more like that of a 100 year old weather-station, but to do that you need to launch the replacement before Kepler fails to allow for calibration for secondary data (such as stellar variation). Now the next few hundred or few thousand exoplanets may not be worth $170m/yr (my guestimate from earlier), but the assumption from others on this thread is that they are worthless. That Kepler’s job was to show exoplanets merely existed, done, move on to a new shiny thing. I think that’s not only wrong, it’s part of a way of thinking that led to JWST and Constellation, and the Shuttle before that, and a culture that can’t develop on time on budget.
Can’t we, just once, reward success instead of rewarding failure. Isn’t that worth the price of admission?
“one Kepler than with twenty five Keplers.”
“25 Keplers” means the replacement program operates over 100 years, since I was talking about one every 4 years. So yes, I’d be happy if we had a facility operated continuously for 100 years.
“Imagine if we bought five SpaceHab flights instead of Hubble.”
Hubble was an oddity, because the primary contractor should have been more familiar with the spacecraft from their spysat development, the original costing/schedule should have been more realistic. Perhaps program managers changed the design too much for the first version, you’d need to ask the people involved. NASA may have been better off ordering multiple Hubbles, starting with one as little changed from a spysat configuration as classification allowed, then adding the new requirements one vehicle at a time.
“Imagine if we bought five SpaceHab flights instead of Hubble.”
Imagine if we’d started with SpaceHab, then developed a free-flier, then a long-duration free flier, then… instead of trying to develop Freedom Space Station in one step. After 20 years and god knows how many tens of billions of dollars, do you think we’d have a facility today less capable than ISS?]
Isn’t Kepler a very application specific gizmo? How do you incrementally go from Kepler to something that does something other than what Kepler does? Are you going to get a spectrum from a transit using a system designed to look at a hundred thousand stars at once? When you already know when and where to look?
If you’re telling me its job isn’t done than sure, that sounds legit. Survey a different part and see if the data agrees, sure.
But doing it again simply because the first one worked is not enough. That it only costs a hundred million or whatever does not mean it’s worth a hundred million.
Never mind the sunk cost.
Unless JWST fails or winds up in the ocean or something, I can almost guarantee we’d be happier with one JWST and one Kepler than with twenty five Keplers.
Imagine if we bought five SpaceHab flights instead of Hubble.
Kepler is awesome and the science it provided truly expanded our view of how many planets could be out there. We even found some earth and super earth planet candidates.
However, I am not sure we should repeat this mission – its time to push the bar forward with a follow on to Kepler that expands on what we learned with Kepler. Whether that is TPF or something else, lets not no “re-do” the mission.
One more thing – Kepler, to me personally, is the greatest SMD mission in my life time. I remember following Kepler back in the 90s when the webpage was that beige color and a bunch of text. You could download the paper model.
I remember when Kepler was not selected for flight and William Borucki wrote a spirited defense of the mission and how he would continue to refine Kepler. I also remember when Kepler was finally accepted and the cheese 16 bit graphic that was placed on the page. It was awesome.
I hope the Kepler team can continue to sift the data and give us more targets to investigate.
Finally, I hope more missions sprout from the Borucki tree and we get these types of missions in the future that inspire the human imagination and STEM in our schools. I hope I get to meat Mr. Borucki someday.
Respectfully,
Andrew Gasser
TEA Party in Space
I must respectfully disagree with respect to “re-do”ing Kepler. It’s not a question of repeating what’s been done, but rather expanding the volume that’s been searched. If we’d put a single rover on the Moon or Mars and let it check out an area of, say, 100 square miles, would we then be justified in using its results to characterize the makeup of the entire planet? If we launched a Sun study satellite and let it observe and record the Sun for one year, would we then be justified in using its results to characterize all the workings of the Sun? Similarly, although Kepler has given us a lot of candidates for potentially being the kind of planet(s) we’re looking for, it’s only looked at a relatively small portion of the viewable universe. The existing Kepler data can not simply be extrapolated and be used to characterize those stars that Kepler has not looked at. To get meaningful data, as opposed to hypothetical assumptions, we really want to do Kepler-like study of everything “visible,” and ideally every observation should be repeated at at least one other point in time.
Not just expanding the number of targets. Kepler’s search requires multiple transits for confirmation. For a Jupiter, that means 15 years minimum. For Saturn, 90 years. So we haven’t even scratched the surface of the targets we have looked at.
Bear in mind that the angular size of a solar type star from the distance of Jupiter is *extraordinarily* small. Detecting planets in Jupiter-like orbits is really, really hard this way. In fact, detecting Jupiters in such orbits is a lot easier and more efficient with spectroscopic velocity signatures.
No, a second Kepler would not be that scientifically valuable. You want to spend $500M to detect the (at last count) 106th extrasolar planet? Kepler was amazing because it established that planetary systems are very common. A second Kepler would just add some frosting to that cake. What is needed is a mission that goes beyond Kepler, ideally actually seeing these planets.
On that topic, let me point out that a second Kepler will NEVER see any of these planets directly. JWST might. (Kepler won’t detect any primeval galaxies either …) So before we’re so quick to declare JWST a failure, before it ever sees light, let’s just acknowledge that programmatic mistakes were made in it’s costing, and move on.
I disagree. Comparing Kepler and JWST really has no meaning; they are two very different systems for doing different jobs. And the differences in build and operational costs make a comparison invalid. What Kepler has been doing, at a very reasonable cost, is of value, but is not completed, even though the original life cycle is done.
I would like to see Kepler’s job continued, without enhancements or changes. Operate the existing Kepler for as long as it continues to generate new data and/or provide confirmation for detected candidates.
If Kepler operation degrades further, I think a Kepler-2 to carry on would be a good investment. I don’t consider it reasonable to take the Kepler data to date as representative of the whole universe, or even our galaxy. That sort of simple extrapolation is not science, it’s unfounded assumption. There’s no substitute for hard data, and Kepler seems to be a cost-effective way to get this particular data.
… in response to Whitfield
“Comparing Kepler and JWST really has no meaning; they are two very different systems for doing different jobs.”
That’s exactly right. I never said that wasn’t so. What I was doing was responding (somewhat sarcastically) to the OP, who was doing precisely that meaningless comparison.
That Kepler results can be taken as representative of other parts of the galaxy is, for lack of any assertion why other parts of the galaxy might be different, a pretty reasonable assumption. If we believed that might not be the case, that the distribution of planetary systems was highly localized, it would have been far smarter to have Kepler do it’s mission split in a couple of different fields. In fact, the Cygnus field that Kepler is looking at can, by that same assertion, be considered not to tell us anything about our own solar neighborhood. On the contrary, we’d like to believe it probably does.
Nope, a Kepler-2 isn’t worth it, if it’s just about pointing in a different direction. It’s worth it if we don’t have better things to do, but we do.
Replying to Helen,
What if the first CfA Redshift Survey had been considered good enough, no need to do CfA2? As it turned out it was the broader second survey which revealed the surprising large voids that make up the universe, which from what I have read no one at the time was predicting as being likely.
I realize that CfA is an imperfect comparison with Kepler since different techniques were used for CfA and CfA2, whereas this discussion is about doing “more of the same” with a Kepler 2. But the point is that until we gather broader data we won’t know for sure if what Kepler has revealed so far is a reliable statistical indicator of what exists throughout the galaxy.
I realize that you are not saying that a similar surprise is not possible with a Kepler follow on, only that you feel that (paraphrasing) unless someone can assert why other parts of the galaxy might be different we have better things to do at the moment (sincere apologies if my paraphrase is not what you meant)
You mentioned an interesting alternative that could have been done with Kepler which is to look at different fields, but you again qualified doing that with a need to have a reason to believe that the distribution of planetary systems was highly localized. I guess that gets to the heart of the differences of opinion on this subject.
Understanding the distribution of planets in our galaxy seems pretty high priority to me. You mentioned that a higher priority at least for now is to know more details about the planets which have been discovered so far. No disagreement there if we had to choose one or the other. But hopefully we can do both. Of course that’s easier said than done. And you are much more knowledgeable than I am of what it takes to justify a mission to the people who hold the purse strings, and I wouldn’t be surprised that unless “assertions” and “beliefs” are put forward it would be very difficult to get funding for a Kepler 2 just because we think we might find something surprising. Unfortunate, but probably the reality.
“I realize that CfA is an imperfect comparison with Kepler since different techniques were used for CfA and CfA2”
No, that’s not it at all. The CfA Redshift Survey was ALL ABOUT spatial structure. Understanding the large scale distribution of galaxies in the universe. So looking in more directions is EXACTLY what strongly enhances the whole effort.
Kepler is NOT about the distribution of planets in the galaxy. Go look at the defining scientific questions for the mission. That question isn’t one of them.
So looking in more directions just adds incrementally to the main conclusion of Kepler, which is that planetary systems are very common.
But seriously, unless there is a fundamental question about the distribution of planets in the galaxy, looking in another direction is just makin’ stuff up as you go along. BTW, Kepler samples a pretty small chunk of the galaxy. What’s the hypothesis you’re testing? Is it good science? OK. Is it important science? Well, maybe. Is this all about chasing serendipity? In fact, if that question were important, it should probably addressed first with ground based RV surveys. That’s a strategy that is hugely cheaper, and vastly less sensitive to orbital inclination.
Again, if I had $500M to spend on exoplanets, I wouldn’t build another Kepler, as marvelous as it was. The NASA ExoPAG, which is constituted to think about precisely this question, has a whole range of study reports on strategic missions that would greatly advance the field and answer important new questions.
I think my own reaction wheel on this thread is exhausted.
Is there a point? I think Kepler pretty much nailed its mission.
We have enough data to make generalizations about the galaxy; to say it may have such and such number of terrestrial planets in habitable zones.
While a fleet of Kepler clones could certainly rack up a list of tens of thousands more planets, I don’t know what we’d do with that list right now.
Can we make better observations of transiting HZ planets? Can we observe their atmospheric gases during transits?Would an orbiter be better at that than ground scopes? Would a new sat be necessary or can Hubble pull off that sort of thing?
It would be a shame to see Kepler go offline soon. The observations it made and inferences from them helped to not only validate transit observations previously made from earth-based telescopes, but pushed the ability to observe much smaller earth-like rocky planet transits. I agree that at this point, additional Kepler type missions will only expand the catelog of likely panets outside our own solar system and from a statistcal standpoint, power a better prediction of just how many planets may be out there. But the essential question of ‘are there any earth-like planets out there’ has been mostly answered now, which was Kepler’s main purpose, imho. Thanks NASA team for a great tool!
Like others, you seem to be assuming that the universe, or at least our galaxy, is made up of a uniform distribution of planet types. This is an assumption only, and to the best of my knowledge, we have absolutely nothing in the way of either data or theory to support that assumption. It was assumed for years that the hidden far side of the Moon was essentially the same as what we could see on the near side, until we actually got to look and found out that it was quite different, for reasons which are still really only theoretical. “Expanding the catalog” is not just a matter of how many planets; the positional distribution of various planet types within the galaxy, for example, is also important data. We may find out that the galaxy also has a “far side” with very different characteristics, and that would open up a whole new list of questions.
Kepler has earned back its cost in droves. The data gained will take years to review but the knowledge is priceless. I’m not convinced JWT will produce what’s intended and prove its investment was worth it. Kepler has done that, in droves.
Damn Aliens!
Always trying to keep us human folk down!
Q: Exactly how many “Kepler” missions would the JWST buy?
A: Somewhere in the ballpark of 16-18.
Is there truly any greater display of the magnitude of JWST’s failure? Kepler is exactly what space science missions should look like. JWST is it’s nighmarish anti-self.
I’d love to see a few more kepler’s up there scanning different parts of the sky.
Jonathan,
While most of us will agree that JWST has been a major disappointment so far, I think we have to be a little bit more flexible in stating “what space science missions should look like.” In many respects JWST is a first, and with firsts we learn things, often the hard way, that we couldn’t have known beforehand. And it’s generally the accumulation of those things that bites us, rather than one specific surprise. Personally, I can forgive a program team for getting into trouble; it happens. The important things, in my mind, are: 1) learn from it and don’t repeat those mistakes; and 2) the moment you discover you’re in trouble, scream loud and clear and start doing everything you can to get back in control — don’t try to hide it or minimize it hoping that you’ll find some magic way to make it all better; that is the unforgivable mistake. It appears to me (an outsider) that we didn’t know JWST was in big trouble until well after it clearly was. I have to wonder, at what point in the chain of command was the message originally hidden?
Dont forget the rolling snowball effect of:
Start
OMG, this thing is going to cost a LOT
Therefore it must not fail
Send some extra observers & inspectors to “Help” make sure it does not fail.
Return to Start (add “More” to “LOT”)
The last 5% of reliability often results in 90% of the cost…
And not just reliability. The longer you wait to report/correct any deviation from the program plan, the more it’s going to cost.
When people try to hide problems, they only come out anyhow later on, much worse. Then (far too late) management starts throwing manpower at the program and when it’s all over, 80% of the total effort and cost went into what was, according to the schedule, the last few percent of the program.
I believe that in most cases this can be avoided by paying for a little overtime or outside help when the original slip/problem actually occurs. Then it doesn’t get a chance to snowball.
Communication is a big factor in all of this because it only takes one person to report “inaccurately” to put the whole program in serious jeopardy. Human nature is one of the most expensive overhead costs.
In many situations (and typical at NASA), programs are in trouble before they even start because the program plan was optimistic, naïve, buying-the-job, or otherwise understated, either deliberately or on purpose. As long as this goes on cost-plus will continue to exist in government programs and cost/schedule overruns will be the norm.
It is my understanding that it is standard practice for NASA program management to hide the bad news about the budget from the public until the program is too far along to cancel, although I suspect key Congressional appropriators are constantly aware of what the situation really is like.
NASA projects are no longer about developing critical technology or important science; they’re about delivering space budget pork to the appropriate Congressional districts. So when a project goes over, it just means more pork delivered. If Congress was really concerned about overruns, they would insist on gutting NASA managers responsible for the program’s mess, which almost never happens.
Skepticism can be healthy, but sarcasm doesn’t help. Calling something “standard practice” implies that it is sanctioned by those in charge, which is not realistic with respect to your assertion.
but sarcasm doesn’t help.
What sarcasm?? My comments are based on my observations and what I have heard NASA people say in private.
Calling something “standard practice” implies
that it is sanctioned by those in charge,
From what I can tell, it is sanctioned.
It’s a game called Getting Congressional Funding and Spreading the Pork.
More pork = more votes. Thems the rules.
This is what I meant elsewhere about human instinct being against us. Attraction to the shiny new toy, and also the sunk cost fallacy. I agree with Gonzo, except I don’t think it’s intentional. Managers are just adapting to how things work. (They don’t do it well enough for it to be intentional.)
Compare Dragon and Orion. Two different strategies. Dragon starts with a working cargo capsule, and is redeveloped into a LEO crew capsule using the experience gained flying the cargo capsule. Once the crew version is flying, presumably SpaceX would use that experience to develop a BEO capsule.
Orion, otoh, was designed to be an oversized BEO capsule first. It is therefore worthless until it’s finished. It taps deeply into the sunk cost fallacy. If you cancel Orion, you “waste” all the money you’ve spent up ’til now.
But imagine what would be happening now with Orion’s future funding if it had developed incrementally like Dragon. What chance does “slightly improved specialised version of existing thing for a future program that doesn’t exist yet” have against rival proposals? You can cancel the new development without having wasted a cent, because you have the results of prior spending flying in space, servicing the ISS.
How many times do you think NASA managers have to see that pattern repeat itself before they start to see all “step 2…” on proposals as just decoration, since no project will ever get funded beyond “step 1”?
[SEV is a semi-exception. But even there, it’s more a “hanging on to step 1 by their fingertips” than a genuine step 2. Which is a pity, there’s so much potential. There was even a lunar lander variant proposed… and… not chosen ahead of the one-step Altair, because… well, see above.]
[[Semi-exception number two, extending existing programs. This fits the sunk-cost fallacy, because it’s “another $X million” vs “throwing away an $X billion project”. The shuttle program lasted over 25 years thanks to this effect.]]
[[[Shuttle-C probably could have been developed had they flown it before the first crew version, as a “safety test of the shuttle stack”. But waiting until after the crew version meant no other version of the shuttle would ever be developed.]]]
[[[[I get a chuckle when people dream about SLS’s 130 tons. If the SLS block 1A gets funded I’ll be surprised. But if block 2 gets built I’ll eat my mousepad.]]]]
To those who lobby for Kepler 2.
Kepler’s original mission and question was just to prove how widespread(if at all) extrasolar planets are. As a bonus-try to find planets similar to Earth in size and orbit around their Sun. The primary question is mostly covered, and we have started getting answer on the second one. Even if the telescope fails-there is still I believe at least 12 months of data to be analyzed(correct me if I am wrong).
So there is no sense to create Kepler 2. You need a new goal. The obvious one seems to continue searching for another Earth-like planets. The devil in this is what method you want to use and where to look? Personally I would try the closest neighborhood but some might disagree. You essentially need something like redesigned SIM Lite mission.
http://en.wikipedia.org/wik…