So whose assumptions are ya gonna trust more?

byline: Ken

Well, Mr. Bonin’s second part is out, and golly if he doesn’t cover a lot of stuff I had in the post. But if it makes sense, then it’s going to sound similar, isn’t it?

Mr. Mark Whittington, who can sometimes be found hanging out here in the Selenian Boondocks, notes over at the Curmudgeons Corner:

“Grant Bonin concludes his polemic against heavy lift and his case is, alas, unpersuasive. As to why it is, just count the number of times he uses the words “assume” or “assumption” in his piece. Economic analysis based on back of the envelope calculations based on assumptions tend to fall down on close examination.

For a somewhat more rigorous cost justification for ESAS, based on some rather sophesticated [sic] cost modeling systems used by both NASA and the Air Force, download the PDF file on Costs.”

Thing is, Mr. Whittington (Mark) is treading on thin ice. I checked the Bonin piece and variations on assume were in there 8 or 9 times. I then went to the Cost section of the ESAS report and found it over 50 times in the 42 pages. This is of course Section 12 of the report, which itself had a whole section (Section 3) on Ground Rules and Assumptions.

I’m actually kind of interested reading through it, though more than a little bit annoyed. There’s really no good cost numbers for this poor investment analyst to dig through other than some projected (i.e. assumed) performance figures for the beknighted (sorry, I mean chosen. I’ve been reading Mark’s stuff too long 😉 launch vehicles. If there are numbers for $ then they aren’t noted or quantified as such, which I seem to recall is what got MPL in trouble. Are those $Bn, $Mn, $M? (thousands for you non-financial types out there, also $m (for mille), as $M is sometimes used for million)

The Ground Rules and Assumptions section is a little more lively, and quite a bit more interesting.
-We find the long sought after source of Human Rating – NASA Procedural Requirements (NPR) 8705.2, “Human-Rating Requirements for Space Systems”.
-aborts from the Lunar surface will take no longer than 5 days for return, independent of orbital alignment.
-CEV will deliver crew to ISS till 2016
-CEV will deliver cargo to ISS till 2016
-CEV ground ops will be at KSC
(In the cost section they note that Sytems Engineering and Integration is estimated at 7% of total cost with staffing cap at 2,000 persons as compared with 1,000 to 1,500 in prior crewed programs)
-JSC gets the space ops part

-The Study will utilize the “Mars Design Reference Mission (DRM) known as DRM 3.0, ‘Reference Mission Version 3.0 Addendum to the Human Exploration of Mars: The Reference Mission of the NASA Mars Exploration Study Team EX13-98-036, June 1998′”

(I believe that’s the document found here. It’s where the 85 metric tonne [m.t., ESAS uses mT] lift requirement comes from, and since the Moon architecture is supposed to be extensible to the Mars architecture, well, might as well start out big)

-the architecture supports global Lunar access. This will be done via EOR-LOR as is noted later.
-the architecture will support a permanent human presence on the Moon (but is it sustainable, not just supportable?)
-In-space EVA assembly will not be required [!?!]
-Human-rated EELVs will require new dedicated launch pads

Assuming that we’re going to use the same shuttle launch facilities and equipment, maintained, gives your assumed HLLV a huge head start if the EELV automatically has to build a new facility. I’ve seen those crawlers they use. Those things are decrepit and the massive tread pieces are cracking. There’s no way to tell from the numbers presented if new crawlers are priced into the cost of the system but I’m guessing no.

Another interesting tidbit: Foreign assets utilized in LV configurations in this study will be assumed to be licensed and produced in the U.S. (Again, is the cost of establishing those facilities for producing the “Foreign assets” baked into the equation?)

The Summary also informs us that “Initial analyses eliminated libration point staging and direct return [lunar surface direct to Earth surface, I assume] mission options”. The Lunar Surface Activities section doesn’t really tell us a whole lot about Lunar surface -activities-.

There’s a lot of stuff to chew on in the report and I recommend strolling on over to to check it out for your self. Please note some of the .pdfs are pretty big. It’s kind of fun to read through and see where they don’t really want to talk about something because they use the same phrases like ‘too demanding’ or ‘too complicated’.

I’ll be back in a little while after I’ve had a chance to digest it a bit more. I’m wondering about some of the delta-V figures, and why in the Summary they bend over backwards talking about LOI and TEI delta-Vs, as well as plane changes, but not the delta-V for TLI, nor the delta-V to/from the Lunar surface. I’m also working on a future piece about the assumptions underlying Direct Trajectory, which itself underlies a lot of choices in the space field.

The following two tabs change content below.

Latest posts by lunadyne (see all)

This entry was posted in Uncategorized. Bookmark the permalink.

17 Responses to So whose assumptions are ya gonna trust more?

  1. Anonymous says:

    Minor nit with Bonins’ numbers.
    Near the bottom of page one he compares amortization costs of heavy vs medium for the missions.
    He uses $10B vs $1B development costs amortized over ten years of launches. The launches total 50 tons in TMI per year.

    Somehow he gets $97,000.00 per kg and $3,300.00 per kg amortized development costs. With the mediums costing 1/10 as much to develop, the numbers should divide by ten.

    I get 50,000 kg per year times ten years is 500,000 kg total launched over the course of the program. The $10B heavy costs would seem to divide to $20,000 per kg and mediums to $2,000 per kg.

    Even the time value of money(interest) doesn’t seem to clear this up completely.


  2. Anonymous says:

    Whittington uses his blog to take short potshots at people without doing any detailed analysis or commentary himself (or allowing anybody he attacks the opportunity to reply on his website, shades of NASAWatch). But once you know what kind of stripes an animal has, you can figure out what it is.

    As for Bonin’s piece, however, Whittington does have a point that Bonin makes a lot of assumptions. But the real problem is that many of Bonin’s assumptions unsupported; they are not based upon actual data, but upon a theory–he posits a number for various things and uses that as the basis for his assumptions, without any proof that the number he has selected is true.

    He did this in the earlier parts of his article as well. For instance, he made the assumption that there is little difference in cost-per-seat-mile for big commercial airliners and small ones. But that is simply not true. Larger jets _do_ have a lower seat-mile cost (assuming that they are modern; an old 747 is going to cost more because of the aging and maintenance issues alone). The problem is that this is not the only relevant factor. Other factors include how many gates are available, how many passengers want to fly a route, etc. So cost-per-seat-mile is an important factor when airliners buy an airplane, but it is not the only one.

    The problem I have with Bonin’s claims is that although they might be right both in general, and in theory, they need to be quantified in some way. He cannot simply assert that _all_ higher flight rates for medium or small launch vehicles will be cheaper (and therefore better than an HLLV). He needs to provide solid data that shows the costs at these higher flight rates. As noted in a previous comment, launching 8 times instead of 4 times might be cheaper per launch, but if you need to build additional facilities in order to launch more than 8, then suddenly 9 launches could be uneconomical, whereas 8 launches are economical.

    Similarly, he assumes that it is better to have a worker do something multiple times instead of once or twice, so having a launch crew launch 20 rockets is better than having them launch 2 rockets. But what if they are incapable of launching more than, say 12 rockets no matter how hard or fast they work? Or what if they become fatigued when they have to launch more than 10, and start making mistakes? Bonin says that people will be inefficient and make mistakes if they do something too little, but never acknowledges that having them do things too much may also lead to mistakes. There are reasons why factories employ more than one person to do the same task–because one person cannot do everything.

    So his assumptions need hard data to support them, and he doesn’t provide that.

  3. Mark says:

    I find it fascinating to be attacked by someone calling himself “Anonymous.” I did not do any detailed analysis on Bonin’s piece because he doesn’t provide anything to work with. It may be a “pot shot” to point this out, but as the same unknown person points out, it was one that hit the target.

  4. Fred Kiesche says:

    “Whittington uses his blog to take short potshots at people without doing any detailed analysis or commentary himself (or allowing anybody he attacks the opportunity to reply on his website, shades of NASAWatch).”

    Hey, it’s his blog. If you want to argue with him, you can always e-mail him. Same with NASAWatch. I’ve found both Mark and Keith accessible via e-mail.

    “As for Bonin’s piece, however, Whittington does have a point that Bonin makes a lot of assumptions.”

    Aha! So despite being a “short” potshoter, he does score. Maybe short postings get to the point faster?

  5. Mark says:

    Fred. thanks. I always believed in what the Bard said about brevity.

  6. murphydyne says:

    One of the reasons I don’t often get to respond in the comments section is that Jon has to have the spam filter on. Unfortunately, the software rarely recognizes the input as correct, even when I try a number of feasible variations. I don’t know why it sometimes does it and other times doesn’t. No idea. So…I have to inflict what was meant to be a follow-up comment in the preceding thread on the general reading public.

    Well, anonymous, it’s not exactly like there’s a whole lot of data out there to work with to quantify things.

    Unfortunately, y’all don’t pay me enough to do that kind of analysis, though if Mr. Bonin is a student he might be able to conduct it as part of his studies. Good luck to him on finding all the data he’ll need.

    It’s funny you brought up the airplane comparison. Part of my day job is analyzing aircraft transactions. The factors you note are only a few of the things we have to look at, but no matter how you slice and dice it there are always just a few salient factors that will really affect the decision. Those factors vary from transaction to transaction, but not much. But in all cases it has to be considered as part of the larger economic picture, and this is I think where the ESAS report falls short. It’s focused in and on itself and doesn’t really show how it fits into the bigger American picture.

    I can also, after fifteen years in finance and banking, smell a loaded deck when I see one. There was a desired outcome, in my opinion, and the various Ground Rules and Assumptions stacked the odds in favor of the house. Mr. Bonin may have made some assumptions, but NASA did too, and as we say down here in Texas no matter how artfully you make up that pig it’s still a dressed-up pig. (I was going to make a polished scatologic reference, but decided to go PG). It’s nice that NASA used advanced modeling software to refine the cost estimates, but if you’re building in crapppy assumptions then it’s like the computer guys say: GIGO.

    I also have to wonder what kind of assumptions went into the determination that the hypothetical Atlas V CLV would have a mean LOM of 1 in 149 (0.6711%) and mean LOC of 1 in 957 (0.1045%) and Delta IV CLV 1 in 172 (0.5814%) and 1 in 1,100 (0.0909%) respectively, while the shaft CLV has a 1 in 460 (0.2174%) and 1 in 2,021 (0.0495%) respectively. Or put another way-

    Loss Of Mission
    A-V: 0.6711%
    DIV: 0.5814%
    SRB: 0.2174%

    Loss Of Crew
    A-V: 0.1045%
    DIV: 0.0909%
    SRB: 0.0495%

    Remember, these are all hypothetical vehicles. These are all built on assumptions. And I for one am more than a little doubtful that either Boe or Lock would sell (different from subcontracted services, a transfer of the equipment certificate) a crewed vehicle to the American taxpayers that was more than twice as dangerous ‘statistically’ as the SRB vehicle.

    Look, I want America in space. I want American business and American science and American economic prosperity and even NASA in space. Since these are my tax dollars being spent anyway irrespective of my own personal ideologies, I would appreciate it if at least it contributed to a much better economic future for this nation by NASA working with American industry (and not just its old reliable chums) to make it so for all of us.

    I honestly think Mr. Griffin gets this, but he’s hobbled by an apparatus that just does not want to change, and may not be able to in its ossified state. He’s hitting the same wall that O’Keefe ran into and it’s starting to get really frustrating. My opinion is that this is why he answered no this week at MSFC and the audience was just stunned. They could not accept that while Bussard ramjets are cool and all, what we really need are the kinds of refinements to existing and near-future vehicles and rocket motors that will decrease costs for EVERYONE. That is what this nation needs right now and they have to get their brains wrapped around that fact.

    Also, things like common interface standards and performance comparisons (like white noise graphs of different vehicles overlaid on one another to figure out the design solutions that would allow one payload to launch on multiple vehicles) would be the kind of useful stuff NASA could do that would be of benefit to American industry. They probably do it already and no one knows about it.

    I honestly believe that we can use existing and near-future launch capabilities to start pushing out into space now, 20 metric tonnes (m.t.) at a time. It allows the kind of flexibility that allows commercial entry attempts by marginal companies, from amongst whom sprout the future economic titans. It also starts us earlier in building capabilities in space outside of NASA, because companies can and do purchase launch vehicles approaching 20 m.t. right now for GEO sat launches. It’s insurable, it’s about the size of a Bigelow Nautilus, and all the big guys (and several smaller ones) have indicated they can do a Crewed Space Craft in the 20 m.t. range. It’s also a good practice size for larger payloads.

    Orbital assembly is a part of this. It is something that we’re going to have to figure out anyway, now’s a good time to get started. I know there are a lot of folks out there taking good hard looks at all of the ISS stuff that has been accumulated to date. There are a lot of ‘almost-assets’ out there floating around. There’s also a lot of interesting capabilities coming along in the next few years.

    Having more actors in the play means more cash flow streams to draw from, which helps to spread around the expenses. Universities getting alumni funding for space experiments for their faculty (snicker, only if there’s football involved, though it’s worth a shot). Private consortia sending up researchers for multi-month stays in orbit slaving over a lab bench. Equipment companies leasing time on free-flyers. Engineering companies providing on-orbit logistics.

    NASA could devote its efforts to actually figuring out what it actually wants to send to the Moon. A nice Caterpillar rego-dozer would be cool, whoops that contract was cancelled. One commenter on one of the message boards (Space Fellowship, IIRC) posed the question “So how is NASA going to pay for all the hundreds of tons of assets of whatever it is it wants to put up there?”

    Good question. But that’s an assumption I’d rather leave alone.

  7. mz says:

    Really good and informative post and lots of links, the Mars reference mission etc. !
    Keep up the good work!

  8. murphydyne says:

    Thanks mz, and sorry about that first paragraph everyone. I thought I was going to have to post it on the main board, but the spam filter changed and I was able to get one through (though not a snarky follow-up. That’s for ya).

  9. Tom Cuddihy says:

    The biggest problem I have with Bonin’s piece, and in general with the MLV argument, is that universally people making that argument seem to handwave away the question of infrastucture required in orbit for rendezvous ops from MLVs.
    I don’t have a big beef with Bonin’s assumptions of cost–he’s got to start somewhere. But I do have a big problem with the fact that no real attention is paid to just what costs rendezvous imposes on the system (other than schedule related.) The assumption is not even stated, although clearly one of the larger considerations for NASA is the ‘parasitic mass’ issue.
    There’s two ways to go about it, the ‘Depot’ method and the ‘lego’ method, which Bonin apparently assumes.
    A lot of people avoid the depot method because it so obviously requires a monster in orbit–another ISS sized tanking facility, and all the associated rendezvous, docking, and fueling ops from the supply ships. It’s a visible, clear blackhole in such an architecture.
    So most people prefer the ‘lego method,’ which has far more hidden parasitic mass and costs. But there’s no visible chuck of depot in orbit.
    Spacecraft are not legos–and if you want to design ESAS spacecraft as flexible and easy to combine as a bunch of legos, there is going to be a significant portion of the mass in orbit devoted to the compatability, rendezvous, and integration issues.
    People like to disregard the ugly example of the ISS as merely ‘a shuttle problem’. Yet Mir faced many of the same problems with getting stuff launched on time, spiraling costs, and integration issues. Leave aside schedule issues–lots of separate modules leads to lots of separate ways for integration complexity and costs.
    This is rarely addressed, and Bonin pretty much assumes it away.

  10. Tom Cuddihy says:

    Just to clarify–Bonin mentions additional complexity with rendezvous and docking, but handwaves it away with “Since only structural considerations would be primary, the entire assembly and integration process would be much simpler than for projects like the ISS.”
    That doesn’t cut it. It’s an enormous assumption that eats around the major issue of complexity that makes an HLLV an easy decision for NASA to make, and especially for an administrator who understands how program complexity builds on itself.

  11. murphydyne says:

    I’m working on a response Tom, but you make it a bit difficult. To help me out a bit can you flesh out some of the things in your post. In particular, I need some definitions, such as:

    -Infrastructure required in orbit for rendezvous ops
    -costs rendezvous imposes on the system
    -parasitic mass
    -depot obviously requires a monster in orbit (why obviously?)
    -visible chuck of depot
    -significant portion (of mass devoted to CR&I)
    -why is complexity a bad thing to tackle? (Private industry does it every day)

    Just trying to avoid any undue assumptions…;-)

  12. Tom Cuddihy says:

    – that should read “chunk of depot in orbit”
    -the concept of parasitic mass is commonly bandied about:it’s generally the percentage of each individual unit’s mass that has to be devoted to systems required for individual launch, rendezvous, and integration
    -as for how much of the mass and cost has to be devoted to that for each component or unit, it is tough to spell out the details of something that doesn’t physically exist yet. You’re asking me to do exactly what I’m arguing Bonin (and you) sidestep: pick and argue a specific MLV system for accomplishing the goals of ESAS.
    ESAS’s systems are spelled out with words, viewgraphs and pictures, thus it’s easy to take those systems ‘as is’ and pick apart the details. Since Bonin doesn’t propose a particular architecture , as Ed Wright actually does do in his articles for Ad Astra Online, it allows him to argue against HHLV without being constrained by the realities of any particular MLV approach. This highlights the costs of the HHLV method and entirely hides the costs of any MLV system, but it’s not a fair basis for comparison.

  13. Tom Cuddihy says:

    Even Ed Wright makes no actual statements about how many modules would be required to be flown for rendezvous prior to departing for the moon or Mars, what percentage of the system has to be devoted to structures for mating, independent power and nav systems that can also work together, etc.

    NASA makes no arguments (that I’m aware of) about the economics of MLVs vs. HLVs. It’s the ‘doability’ that concerns people like Griffen. Consider, for example, the general architecture proposed by Bonin, where he actually discusses a 25-mT to LEO MLV that ‘would take 4 launches to launch your 50 mT spacecraft to trans-Mars injection.’ This is great back of the envelope stuff–and also crap when examined in detail. Each time you break a module into smaller segments, you steal a higher percentage of the available mass for necessary systems as a result of the realities of power, thermal control, and pure volume vs. surface area constraints (for Mars entry shields, fuel tanks, or habitable volume, for example).

    In reality, it would more likely require 5 or 6 MLV launches to get your ‘single’ lego spacecraft in sufficient shape to make the jouney to Mars. Perhaps even more. Some components, like a nuclear reactor, many HAVE to be launched in a single unit.

    Add to this the peculiarities of government contracting. As any veteran of government space contracting can tell you, the most dangerous, costly, and schedule prohibitive part of any large program, barring a particular technical hurdle, is invariably ‘interfaces.’
    Interfaces designed by different people working on different systems, that have to go together smoothly, and work correctly or doom the whole spacecraft to failure. Poor interfacing has resulted in catastrophes throughout space history, and is responsible for a significant portion of ISS module cost overruns.
    Worse, it’s not just an aspect of poor program management, but regardless of leadership, interface problems tend to accumulate as complexity increases.
    As complexity increases, safety and redundancy has to be (parasitically) increased as well, on each module. For example, if your particular architecture requires multiple rockets firing at the same time for, say, TLI, that means those components have to be designed to work alone, together, and redundantly. And you have to ensure that a failure on one module won’t propogate to other modules.
    Finally, rendezvous has many many ways to go wrong. Let’s not forget Spektr crashed into Mir. DART’s more recent crash doesn’t exactly instill confidence in NASA’s automated docking capabilities either.

    There can be no question that a many-separate-modules system, launched on many separate launchers, cannot but be more organizationally complex and difficult than an HLV launched system, even if the ground infrastructure is many times more complex.

  14. Tom Cuddihy says:


    that last sentence should read “cannot but be more simple, even if the ground segment is more complex.”

  15. Jon Goff says:

    I still think the best way of doing large missions like this on smaller vehicles is to try and minimize the number of separate modules (unless the interfaces are right along natural “split-lines” anyway), and use a lot of on-orbit propellant transfer.

    You don’t need a depot to do on-orbit propellant transfer, or even long term storage. It would be nice to have something that you didn’t have to drag along with you that could keep your propellants chilled while you’re tanking up, but it isn’t explicitly neccessary.

    I personally think the best way to do Mars missions (at least before we have something like orbital microwave emitters that can get you high Isp, short duration flights) would be to use a vehicle that had 1-4 parallel “boost stages” that are dry-launched, as well as a Mars Orbit Insertion stage, the lander, and a return stage. The boost stages would have enough delta-V to do the Trans-Mars Injection burn, and then after staging to do a burn that brings them back into HEO (after which they aerobrake, get remated, and refueled for another shot). The Mars Orbit Insertion stage would be left in Mars orbit (for potential reuse down the road once ISRU propellants are available), then the lander would go down, setup camp, and then come back up for the return flight. The specific details of whether you use ISRU to cut down on the return vehicle size, the lander size, or reuse the MOI stage depends on the preferences of the user.

    The nice thing about such a plan is that the independent spacecraft have very simpler interfaces, the boosters are all identical (and possibly derived from trans-lunar boosters), and nothing needs heavy lift. Not to mention that using this technique, and oversizing the booster stages, you could likely send off missions even during off-years, maybe even getting a decent number of flights per year sent off to Mars. Reusing that hardware at that rate would drop the costs substantially.

    But that’s just how I would do it. That’s quite a bit of handwaiving without even much BOTE calculations (other than some ones I did way back) to back it up, but that’s mostly because Mars just really doesn’t do it for me. I’d rather focus on the Moon since it’s likely to be the first off-world economic hot spot.


  16. Tom Cuddihy says:

    Hmm….the boost stages would have 4.2 km/s to get from LEO to TMI…then, depending on where you did the return burn, anywhere from 700-1500 m/s to get back to elliptical earth orbit (Unless the moon was lined up perfectly for a return to earth trajectory)

    …so your reusable boost stages are already at 5-7 km/s of delta v, say 5.5 km/s–> for a 50 mT mars mission, you’re talking about a very large boost stage, way bigger than your TLI stage)–that’s a lot of fuel you need in LEO…there’s a reason every MTI injection stage seriously looked at is considered a throwaway. At least a TLI stage can be recovered on the return part of the orbit.

  17. Jon Goff says:

    You need to remember that when the booster has separated from the payload, its dry mass just dropped substantially. The extra propellant for the return isn’t really that bad at all. I don’t have the time at the moment to run the numbers, but others who have (like Henry Spencer) seem to think it makes a lot of sense. I think the reason why all the TMI stages in the past have been designed to be expendable is that pretty much everything in the past has been designed to be expendable.

    Maybe in a few weeks I’ll have some time to do some more detailed number crunching.

Leave a Reply

Your email address will not be published. Required fields are marked *