JPL's InSight: Ignoring The Real Costs – and its MPL Heritage
Keith’s note: According to the official web page for the proposed InSight mission to Mars at NASA JPL: “The InSight mission will fly a near-duplicate of the Mars lander that the Phoenix mission used successfully in 2007 to study ground ice near the north pole of Mars. The reuse of this technology, developed and built by Lockheed-Martin Space Systems in Denver, CO, will provide a low-risk path to Mars without the added cost of designing and testing a new system from scratch.” No cost numbers are provided to verify the cost cutting claim.
The highly successful Mars Phoenix is (logically) mentioned as a way to claim cost savings. But when Phoenix was proposed the cost savings from heavy reuse of failed Mars Polar Lander heritage hardware were cited – but never fully explained. If this mission is approved there is no doubt that JPL and SMD PAO will once again try and claim massive cost savings and simultaneously not mention the money spent to develop the hardware for previous missions.
Keith’s update: Gee, that was fast. Spin control has begun. JPL PAO’s Veronica McGregor just tweeted “@NASAWatch MPL was a different design from the 2001 lander.” The University of Arizona’s Phoenix site says “The Phoenix Mission uses the Mars Surveyor 2001 Lander, built in 2000, but later administratively mothballed.” According to the NASA NSSDC entry on Mars Surveyor 2001 Lander: “The mission will be based on the Mars ’98 Polar Lander”. Here we go again, JPL is trying to have it both ways – they want you to accept the fact that InSight uses Phoenix heritage (i.e. the Mars Surveyor 2001 Lander) – but they do not want you to know that Mars Surveyor 2001 Lander was very, very closely based on the Mars Polar Lander design – indeed, modifications to what became Phoenix were the direct result of the failure analysis of MPL – that is how closely they were related.
– NASA SMD’s Cost Overrun Coverup (updated with Telecon notes), earlier post
– Yet Another Mars Phoenix Cost Figure, earlier post
– The Actual Cost of Mars Phoenix is $520 Million, earlier post
– Why Does The Official Cost of Mars Phoenix Keep Changing?, earlier post
– NASA Has a Problem Calculating – and Admitting – What Space Missions Really Cost, earlier post
Whatever money was already spent, while it may be the “real” cost of the system, is irrelevant when deciding between the three choices going forward. The missions should be evaluated based on the cost from now to mission completion.Â
The larger problem I see in this, is the constant use of “heritage” in all NASA programs, which often has lots of associated problems that are glazed over, and rarely comes in at the quoted low low price.Â
heritage is really the personal knowledge and experience of the people associated with the design and use of the component, assembly, or what-have-you. If you claim “we’ve got a spare from a previous mission”, but don’t have the people around who were aware of the basis for the design, and it’s peculiarities, you probably don’t save much.
The reality is that there isn’t as much commonality between missions as one might hope. It’s not like building cars on an assembly line, all of which will be used in pretty much the same way. Sure, we send lots of stuff to Mars, but it goes to different places on Mars with different environmental conditions, it gets used to do different things, etc. So the designs and testing are different each time. As another poster commented, test is a big cost, and you have to do that each time, and there’s never enough money and time to do as much test as you’d like.
For infrastructurey kinds of things, there *is* a lot of reuse: radios, batteries, flight computers. But the science instruments are different every time, and that drives the physical layout and accommodations on the spacecraft, which drives things like antennas. THere’s also a tendency to push right to the performance limit, and subtle things like “in this particular launch opportunity, the atmosphere won’t be quite as thick, so we can’t land quite as much mass as last time”.Â
And of course, since software is a HUGE part of the system cost these days, a lot of that does get reused. There’s still a lot of new software each time, because while the spacecraft are superficially similar, they are getting more complex operationally. A lot more autonomy is expected than, say, 10-15 years ago. Â
The reuse of design, and the application of failure analysis, from previous missions I see as definite moves in the right direction. Consider how many past NASA programs have, even at JPL, essentially started from scratch with their designs, all but ignoring what others had learned from their prior experience.
On the other hand, I think, Keith, that they can actually have it both ways, because they are hopefully reusing design concepts, methods and analysis from both Phoenix and MPL. It is maybe a subtle point, since a lot of people seem to miss it or dismiss it out of hand, but reusing design (methods and elements) is not the same as trying to reuse major hardware elements. The same logic applies to processes. The processes used at either end of a mission — 1) build a stack to launch, and 2) “land” on a planet or “rendezvous” with an in-space destination — vary according to requirements from one program to the next, but the logically reusable process elements are re-employed whenever practical and possible.
As I see it, NASA has a bad habit, designwise, of extremes; they either reinvent the wheel across the board on a program, or they try to reuse major hardware components as they are. The former is ineffective and costly, with no synergy, and the latter inevitably ends up with major unforeseen integration problems, and again ends up being ineffective and costly, sort of like the way Microsoft writes software — if you add enough connecting pieces into the path you can make the thing run, end to end, but the path ends up being much longer (and therefore less reliable) than was ever necessary. If ABCD are the four corners of a square, and your goal is to get from A to B, then go from A to B; don’t travel ACDB (a longer path, backwards around the square) just because an AC and a DB were already designed for something similar. For example, the Congress-mandated legacy hardware elements make SLS a “backwards around the square” design (certainly not its only major fault).
If the “reuse of design” concept had been rigorously applied over the few decades at NASA, then we could have actually implemented the many bootstrap, modular and incremental plans that so many of us have proposed, and benefited from all of the synergy and progressive learning that would have accrued.
Steve
Ask them what InSight costs. Â Then wait a few months and ask them again. And so forth. The numbers you get from them are never the same because they just make this stuff up every time someone asks. Truth is – especially at JPL – they really have no idea what things cost – that is why MSL over ran as badly as it did.
Keith,
OK; Apples and oranges. I guess there are two separate issues: 1) whether designing and costing based on previous programs is valid; and 2) whether NASA (JPL in particular) costing exercises in any way reflect reality. In rereading our comments, it appears to me that I was addressing #1 but you were discussing #2.
I don’t think anyone would disagree with your stance on #2, Keith. But, then again, we all know the reasons why program costing is so inaccurate, and why program costs change so radically over the life of a program.
Also, if you leave aside cost-plus contracts, then, on average, industry’s track record is not much better.
Everything taken together, it’s rather silly that #2 continues to be a problem that is not being rectified, and that the magnitude of the problem continues, on average, to get worse instead of better. I almost have to believe that people have chosen to let the situation persist, rather than invest the time and energy needed to overhaul the status quo.
Steve
Disagree. They are thrashing to get around problems with qualifying the spacecraft by rehashing subsystems until things “work”. They discard or rebuild subsystems that “get in the way” simply because saving the cost of them isn’t the issue, completing the spacecraft is.
I’m not sure you can tell in the middle of this whether its bad planning or simply endemic to making a new class of spacecraft that that’s how things work out. Because of the “gotcha factor”, it is unlikely anyone keeps score to tell why.
You get all kind of weird materials/thermal/environmental issues that you don’t predict that means you fail in thermal/shaker/vacuum/other tests. These are the cost of the game period. Others are stupid, where someone didn’t notice something obvious like mechanical interference.
These affect costing. One problem with costing models is that it is easy to distort them. And, sometimes you get asked to redo the costing model from somebody else’s favorite perspective, often to justify a political agenda, so you end up with too much information saying different things, all true.
Costing isn”t like physics where there is just one true answer – its not like a vending machine with fixed number of coins for a thruster or a transponder …
Back in the days when Mars 98 and Mars Polar lander (01 lander) were designed, Lockheed had at least SIX spacecraft in design/build that all had almost the exact same avionics architecture and the same software (mars 98 orbiter, 98 lander, 01 orbiter, 01 lander , Stardust, Genesis).
Every one of those spacecraft used nearly identical hardware at the core, so the software was reusable even though the missions were vastly different.
This allowed a huge savings in $$ and development costs.
The sad part of trying to reuse old proven designs is that it is getting increasingly difficult to find the exact same space qualified electronic components year after year.
Parts obsolescence is your biggest enemy.
Once you have to change the design to accomodate a handful of new parts, it is way too easy to decide to change the whole design.
Example: the spacecraft above nearly all used RAD6000 single board computers, but you cannot buy those any more.
So you couldnt even build exact replicas of the MER rovers without using newer parts (they used the RAD6000s as well)
The answer to the parts obsolescence issue is ongoing finance to retain the qualification of a bus for subsequent missions as parts come in/out of vogue. We do this elsewhere in aerospace.
What the issue is, is stupid financing that is used to reviving dead buses, by conjuring the corpse to get a little more life out of it, rather than forward financing knowing you’ll be using it again. The past reason was that we never expected reuse to matter, but increasingly it does, and that’s why we need to think smarter not more cynically about this.
 I can verify that. I was on the software IV&V team for Mars 98 and Mars 2001 (and although not officially for Stardust and Genesis… since it was the same software…)
If we are going to reuse older technology, how about launching near-identical copies of Spirit and Opportunity? Those missions performed so spectacularly well, new missions based on those would be lower-cost and virtually guaranteed to succeed. (BTW, I said “near-identical”, because it might be worthwhile to fix the few problems those missions did have, such as implementing some kind of really simple way to keep the solar panels clean, a tilt mechanism for the solar panels, and maybe making the wheels a tiny bit more robust)
Oh that will never happen since it makes complete sense to do what you have suggested….
Exactly what I said at the time. With certain refinements, they’d outlast nuclear rovers.
sadly — it would cost a significant portion of the original budget to build another set of MERs. Most of the cost is in test, validation, documentation, etc., which they would redo, right or wrong
What I pushed for (and “no one” wanted to hear) was to change the model for unmanned exploration such that reflying missions would fall under a different rubric than novel ones.
This is a difficult time for such – we are at the stage where we have a number of successful probe buses that can be highly reused for Scout class missions. I think the reasonable fear of optimizing them is that moderate / flagship missions may suffer, especially with cost overruns.
But here’s the potential advantage – create an economic model whereby you have a streamlined test/validation/qualification/encapsulation means through a single vendor with a commitment of N missions over a decade. Then, the various university/other research labs instrument the bus, fit their test/qualification processes into the template, and flow through the master process. This and potentially cheaper LV (these are all Delta II class) through Falcon 9 /Antares might mean you can have half-priced missions, possibly a queue of them ready to fly for the next opposition.
Its no different than the bulk EELV purchase for ULA thats underway.
This is how we need to think about unmanned exploration now.
NOoC,
This is the sort of thinking that I really like. Standardize the process as well as the hardware (assuming that there are no unions or existing contracts that will object). You’re reducing the price while actually gaining capability and performance. Smart.
Do you think that the same sort of approach can be applied to larger flagship missions? Then we would have a standard family of off-the-shelf size options, from cubesats up to large flagship structures. I sort of envision two flagship-sized standard structures, a lander/rover version and an orbital version. The orbital version could perhaps double as a long-range probe like the Voyagers. With this approach, LVs and things like farings can potentially become more standardized as well. (I’m deliberately ignoring SLS.)
If a standard flagship mission structure is reasonable, do you think there is potential to have them include (or optionally include) follow-on facilities such as: standard racking for in-space repair (or even on-Moon/-Mars repair); standard docking ports and perhaps the ability to pressurize an interior volume (for spacecraft docking or ISS docking); plumbing and optional external tankage, so that a probe/lander/rover launched into orbit can top up the tanks before leaving orbit? I’m sure you get the idea. And with the number of (and current love for) larger LVs, we’re going to have the horsepower to lift these non-essential but valuable extras. We might even see a real standardizing of LV list class.
I’m basically thinking in terms of making mission vehicles more capable and more flexible (while still being standardized), multi-mission capable, longer-lived, etc. We lose so much now because we use everything once and then throw it away, then design something completely different the next time, over and over.
I believe that we need to stop doing one-shot programs and start thinking seriously in terms of 1) standardizing and 2) building equipment that will aggregate into the space infrastructure that we have recognized as necessary all along. There’s no way that any one program is going to produce that infrastructure, or even significant parts of it. We’ll have to collect it one or two pieces at a time, so perhaps with “a family of size options” in our spacecraft and their accessories we really can start treating them like Lego and build up that infrastructure. If we take the same approach with bases, space stations and any other in-space facilities, that would hopefully keep any unique items to a minimum. The other aspect with standardization, of course, is that (ITAR willing) it becomes easier and more cost-effective to incorporate international participation, both for planned programs and for any emergencies that develop. It might even make an improvement our horrible record for program costing.
It would all start with your proposal for standardizing the bus and processing for scout class missions and then applying the same logic on a larger scale. Does this make sense? Can you see it as a way to both save money (and time) and make progress towards a space infrastructure without sacrificing performance or capabilities? What do you think?
Thanks,
Steve
Man, I’ve got to stop writing such long posts!
Can you see it as a way to both save money (and time) and make progress
towards a space infrastructure without sacrificing performance or
capabilities?
Everything I’ve worked for has been to establish space infrastructure. Even when I talk of why things are the way they are, its as a means of addressing how to get this into action.
I believe we are on an inexorable course for such. The industry fights this out of past biases, but I think they’re worried about getting any business at all given whats about to happen to SMD – they might listen this time.
You start out with the smaller missions with a bulk arrangement. The reason the larger ones are harder, is because we’re still proving them due to lack of frequency to fly.
Economics are driven by careful use of frequently flown buses/instruments/subsystems. But then you have to forward develop them to keep them vital. To some this is paying for nothing. It’s actually the reverse.
no one of consequence,
Thanks, I’m reassured. Although we may have outlived that wonderful Chesley Bonestell/Collier’s vision of space, we can still believe that Wernher von Braun was right all along.
Steve
The tools don’t exist any more, the people dispersed. It’d be a different team trying to copy existing designs, rather than just ordering up 10 more.
Would have been different had they continued right away. But I suspect it’s been too long.
Of course, knowing that, it would seem like common sense to learn from your mistake and ensure that successful programs can be built on… but that doesn’t seem to be the case. (Where’s MSL II and III, along with the Lunar variants, should MSL be a success.)
(Hmmm, MSL I Curiosity. MSL II & III… Inquiry and Investigation? Schrodinger and Satisfaction?)
[Edit:
“maybe making the wheels a tiny bit more robust”
http://www.youtube.com/watc… ]
 Paul,
Look into the MAXC rover… an attempt to do same. Hope it is revived.
It is a tried and tested aerospace practice to account for the costs of reusing the same satellite bus, especially the case in communications sats. Clear communications of that fact makes it easier to “sell” new missions.
It is likely that future orbiters and one-shot landers (not rovers) will get funded off of high amounts of common bus reuse. I consider this is more due to inexperience and lack of skill than deception. The PAO side gets its wires crossed between the Hollywood styled “never in the history of …” buildup and its true story of reinvestment in common landing/orbiter instrument buses.
Mars probes in particular are a hard area to build expertise. The ability to reuse successful EDL components alone is a uniquely valuable asset.
As to accounting, the “having it both ways” I suspect is due to accounting for the failure lump sum total wrecks the incremental successive mission numbers. This is due to the “sunk cost” nonsense that has no accounting reality but all too great a political reality in Congress with everything NASA.
You seem to forget two things :
1) the main contractor (here LM) will charge you less if it reuses previous design : this help keeping the cost cap when you’re tight
2) In space mission, you have to *qualify* pieces of hardware to make them fly – the TRL story : look on the web- So reusing *qualified* hardware and avoiding as much as possible to develop/qualify new subsystems is an obvious choice