Yesterday, I started to present a framework to think about how proliferators get the know-how and industrial infrastructure necessary to make weapons of mass destruction. This framework has proved very helpful to me in understanding various WMD programs. It becomes more predictive when combined with two competitive forces: the need for proliferators to minimize risk of failure—something all industrial projects have in common—and the need for secrecy. Minimizing risk of failure can be most easily accomplished by utilizing foreign assistance from people who have already produced WMD, or something very similar to it, as much as possible. As has been pointed out by many studies on the transfer of civilian technology, this is the best way of acquiring new industries. Unfortunately for the proliferator, he usually wants to minimize the number of foreigners who know about his programs. Enter the proliferation profiteer. A. Q. Khan is the best know example of this but such profiteers have played important roles in a large number of WMD programs.
What Proliferators Want: Minimal Risk of Failure
Many things can go wrong when embarking on a complex technological venture, which almost by definition emerging proliferators have little or no previous experience with. This is especially true of independent innovation where you could very easily spend billions of dollars and not end up with a weapon. Consider the dismay that occurred when the Manhattan project first discovered how radioactive plutonium was; it was not at all clear if plutonium was going to be useable for a weapon. This is well documented and dramatized in the first chapter of Critical Assembly. If you are going to spend billions on a project, you want to minimize the risk of failure. The Manhattan Project minimized risk by running so many development projects in parallel. (I think this parallelism can go a far way in explaining the Iraqi Supergun project.) It is also why the United States, the country that first invented the liquid propellant rocket engine, shoved aside its indigenous missile development program as it grabbed Germany’s rocket scientists and as much of their production plants as they could at the end of the Second World War.
It’s not clear that proliferators always think this through, however. The temptation to try to reverse engineer something appears to be too great and of them attempt such projects before they have fully assimilated the production technology. Both Iraq and India (see the discussion of India’s Project Devil in Raj Chengappa’s Weapons of Peace ) tried and failed to reverse engineer the SCUD. In fact, I believe that both had difficulty simply understanding how the missile functioned much less reproducing the production line needed to build it. (But that is another full posting all together.)
The Problem(s) of Secrecy
When we think of secrecy and WMD, we usually think of the efforts to prevent the spread of the key secret, such as the chemical formula for VX or the number of neutrons released in a fission. That, unfortunately, misses the point and more than a wiff of jingosim. Those sorts of secrets will always get out and really don’t represent a key problem for proliferators. Their major concern with secrecy is to prevent their activities from becoming known to the outside world. After all, the main reason we know about the Libyan nuclear program is because A. Q. Khan’s shipments were detected by the international community. The dilemma for proliferators is, if they don’t use foreign assistance, they dramatically increase the chances of program failure.
Schmucker’s Ideas, the MTCR, and this Framework
Unfortunately, too little of Schmucker’s ideas have been published in English. If I was to summarize his main points (and I would welcome corrections on this from him or his associates) they include the following (again, I am sure I am simplifying here):
1) Reverse engineering is hard.
2) North Korea has performed far too few flight tests even for a development path based on reverse engineering.
3) When Schmucker reconstructs the development timeline for each missile, it seems far too short for a realistic development program.
4) North Korea does not have the visible industrial infrastructure to sustain a missile development program.
5) There is an uncanny resemblance between the missiles “fielded” by North Korea and obsolete Soviet missiles. (I don’t want to get into the discussion some wonk-readers have had about obsolete vs. abandoned designs.)
From these observations, he draws the conclusion that the North Koreans are getting their “missiles” from Russia. I think he makes a compelling case. For instance, the similarities are simply too great to believe otherwise. (See my post on the Safir’s second stage turbopump, for instance.) However, in the framework I have been discussing, this does not imply incompetence on the part of North Korea. Instead, it simply means that they might have followed the path to acquisition of missile technology that has historically proven to work the best. The lack of visible infrastructure might also imply that the DPRK is not making much effort to assimilate the technology. Iran seems much more interested in that. However, it also does not mean that the North Koreans are simply purchasing missile components. I think the most likely scenario is that they are purchasing “production lines” as well as the know-how to make the missiles. Know-how is a well defined commercial commodity and is purchased daily in joint ventures. It can include things as simple as documentation or be as extensive on-site training for the shop-floor workers.
Finally, let me just say that selling a production line piecemeal (even if it is to the same buyer) is very different than selling missile components in terms of getting it past a country’s export controls. Both might be disallowed under the MTCR, but at least milling machines are dual use. And it is much, much simpler to reverse engineer a production, where you can just order another copy of a specific milling machine, than it is to reverse engineer the missile.
This series of posts was supposed to end today with a discussion of Iran’s state of knowledge in an industrial sector adjunct to missile development: advanced composite material. Ending it today would have allowed me to disappear out of internet contact for a few days. Since that is going to happen no matter what, the last post in this series will have to be postponed until sometime next week.
Thanks for putting up with these long, theoretical posts! I have some ideas for shorter, more fun posts starting next week.
This series of posts consists of:
0) Do You Know What This Thing Is?
3) The How of Proliferation, Part 1
These posts have been fantastic – thank you! And the VX molecular structure is total nerd bait for a chemist like me.
This is an obvious point but may help to simplify/clarify the architecture of proliferation you are analyzing in this interesting way. As set forth in an old book I wrote for Brookings in the early l990s (Trappings of Power), states generally have developed ballistic missiles using one of three principal acquisition strategies: 1) as spin offs of civilian space programs (India); 2) through programs emerging out of dedicated defense industries involved in production or co-production of non-proscribed weapons with the assistance of outside suppliers (South Korea); or, as is increasingly common, through all source shopping sprees – the Walmart model – best typified by Iraq. In this last case, technology and know-how are acquired on an all source basis whether or not the industrial infrastucture exists to apply such inputs in a productive or certainly cost effective way. States with more advanced technical bases typically enjoyed robust defense trade relations with the advanced industrial suppliers and axiomatically have had more chance of success developing operational missile forces.
Janne;
India, as one of its rocket scientists said, had the best missile program money could buy. By that he meant they got started by licensing technology from France and went on from their with full assistance from other European partners. India definitely falls in the “turnkey” category.
Project Devil was an independent military program started well after the civilian program was up and running. For some reason they tried reverse engineering and ran into trouble (the same trouble the Iraqis ran into with their first attempts at modifying the SCUD). Perhaps not surprisingly, the civilians, when they were called in, essentially made fun of the reverse engineer plan and certainly did not appreciate the difficulties inherent in that path toward acquisition.
A little addition on Schmucker’s ideas:
He also puts emphasis on the issue that classical “rev-eng”-products typically differ significantly from the original, either in optical appearance and/or in general performance and/or reliability (up to the point of non-functionality) due to the fact that usually (unless we’re talking about really simple products – the level of complexity of the to-be-copied item defines the level of difficulty of the copying process) not all neccessary information on the to-be-copied item can be sufficiently determined by analysis alone (especially from only a limited number of specimen available for that purpose – and we shouldn’t forget about the financial and operating expense associated with that type of testing to death, either) and thus has to be complemented by engineering solutions known/available;
The same applies to materials: nationally unavailable types have to be either procured from abroad or supplanted by home-grown ones – either with a non-insignificant loss in optical appearance, performance or reliability due to the fact that additional security-margins have to be implemented due to less than full understanding of the involved physical models leading to the exact dimensioning of the original part (e.g. supplanting a 1/8”-sheet with a 3mm-one, even if exactly from the same steel-alloy, will result in a potentially unacceptable loss in reliability, while using a 3.5mm-sheet instead will add weight and thus diminish the performance – and the less margin of safety the original part had, the more crucial ‘overdimensioning’ will become in this context – recall that we’re talking about ‘rocket science’: a rather weight-restrictive, but at the same time extremely strength-relentless application!);
(BTW, if i already have all neccessary knowledge, equipment, materials etc. in full detail, then why should i copy somebody else?)
A very close copy with comparable levels of performance and reliability is thus highly suggestive of considerable help (this can also be of involuntary character, e.g. industrial espionage or forced labor by foreign specialists -> see operation ‘ossoaviachim’) by the manufacturer of the original on the product itself and the production process (and, in case of missile systems, of course also the missile complex, too – a Scud-B without an appropriate TEL/MEL, fuel-transporters or testing-vehicles etc., etc, etc. is far from being operationally deployable!).
As for North Korea, just to show what might going on behind the scene; here something that looks like operational North Korean Nodongs on their modified SCUD TELs.
http://img12.imageshack.us/img12/8136/14154586200706120937305.jpg
Iranian and North Korean levels of knowledge and existing infrastructure should really not be underestimated or compared to countries like Iraq.
I have one suggestion for any future presentation of this framework. You may wish to describe the two considerations to the proliferator as (1) risk of failure and (2) risk of detection. I think “risk of detection” describes your concept better than “secrecy.”
Pedro, thank you for linking to that (quite spectacular) picture, but again, converting trucks is not exactly rocket science (so we shouldn’t overestimate the North Koreans or the Iranians based on such an acquirement). Besides, as far as i understand this thread, it’s primarily not about what the North Koreans are capable of now (or will be in the future), but what they were capable of back in the 80ies/early 90ties, when they seemingly assimilated parts of the old soviet missile program (and i think that process is anything but well-understood). In this context, pre-Desert-Storm Iraq is an excellent example, as is the early indian missile program, both nicely illustrating the difficulties of trying to “reverse engineer” missile technology with insufficient help from the original manufacturer.
BTW, those five-axle TELs do remind me somehow of a crossover of a 9P117M1 and the chinese Wanshan WS-2500 (see e.g. here: http://www.sinodefence.com/strategic/missile/df21c.asp)…
@Jochen Schischka
My point was only that their programs are quite serious. At the moment you and others have even problems to believe that the TEL/MEL’s both countries are producing are for a show-off to the west with no serious operational capabilities nor significant numbers.
The article is very interesting yes and this point of mine might be a too small detail. However we have seen this NK Nodong TELs in numbers while they never presented any of them, not even Nodongs on any parade. Additionally NK has shown 8 or 10 TELs of their new solid fuel short range missiles and Iran 6 identical and operational Shahab-3 MEL’s years ago.
I’d say Forden’s axis of local knowledge and engineering capabilities is quite high for NK and Iran and even higher for China. That’s why they manage to do reverse engineering successfully; maybe Forden can give us an insight into at least Iran’s technological base.
The NK Nodong TEL is most likely a modified, reverse-engineered Maz SCUD TEL. They should have the Industrial capabileties for that.
Pedro:
“That’s why they manage to do reverse engineering successfully”
Can you please give just one single example of this (independent) “successful reverse engineering”? Obviously you don’t read what i’m posting – otherwise you might notice that all these “rev-eng”-products are clearly distinguishable from their originals (either by optical appearance, loss in general performance or lack of reliability), absolutely unlike the north-korean and iranian Scud-B weapons systems, unless there is a high degree of help by the original manufacturer involved (“turnkey-factory”, maybe even only simply import – since those missiles obviously were exported by the DPRK without any type of quality-assurance flight-testing, and the Iranians didn’t have any problems with these during GW1)! And then, i’m asking me how a country (the DPRK) with a GDP less than e.g. Uruguay or Turkmenistan can finance buying complete missile factories or large numbers of missile complexes.
Look for example at the chinese “rev-eng”-R-14/SS-5/Skean, the DF-3/CSS-2 (since you chose to ignore my previous examples of the US-redstone, the soviet Tu-4/Bull or R-1/SS-1a/Scunner, the iraqi or the indian rev-eng-efforts): Less range, other dimensions, no conical tail and four independent rocket engines, each with its own external fuel line, instead of one four-chamber engine with a common turbopump – only on very cursory examination could somebody mistake the one for the other!
“The NK Nodong TEL is most likely a modified, reverse-engineered Maz SCUD TEL. They should have the Industrial capabileties for that.”
On the “reverse-engineering”-part: see above, simply too identical in every detail for a real rev-eng-copy – more likely an imported, at best license-produced MAZ 9P117M1!
But i agree, almost everybody on this planet should have the industrial capabilities to convert something as simple as trucks. The average repair shop should be able to do that, but do you earnestly want to suggest that every car mechanic can design and/or manufacture functional ICBMs?!?
And unless you can put photographic evidence for your “six identical MEL”-claim on the table (which you obviously can’t! It won’t get more true by simply reasserting that over and over again! I’ll be happy to get convinced by traceable evidence, but ad-nauseam-rethoric won’t work!), i’d suggest to stop affirming that.
The Russians used to have two production lines for Scud-type missiles. It is well possible that the Norks purchased one of these. But even if they did so, that doesn’t mean that they are able to mass-produce Scuds – if you are not familiar with the specifications of the materials or the details of the manufacturing process, you might well be able to produce something that looks like a Scud engine – but you would want to be pretty sure that it works like a Scud engine, too. Jochen Schischka has already highlighted the difficulties involved in reengineering – they would almost certainly still apply even if the Norks had the necessary production infrastructure. From what I understood from my conversations with Schmucker, his point applies mainly to North Korea’s capabilities to indigenously develop anything that comes close to an ICBM. They may try to combine existing concepts and pile them up to a three stage rocket, but they are far from developing anything on their own. If one agrees to Schmucker’s theory, another often-claimed link is called into question. If the Iranians are able to place a small satellite into an orbit and manage the delicate process of stage separation, why should one believe that they are technologically dependent on North Korea, a country that has yet to master these milestones in rocket science. Additionally, the whole threat assessment with regard to North Korea’s missile program would warrant reconsideration, if Schmucker is right.
The subject of reverse engineering is very interesting. AQ Khan provides many examples of the dilemmas facing an engineering copycat. If you look at KRL’s work on pressure transducers, you see that Khan faced a choice between making a copy in a hurry of US-origin equipment to assist his enrichment program (he was at that time expanding cascades at Kahuta and he had to get enough of these items to meet his requirements) or to make equipment good enough to find other potential buyers. At this time, Khan was venturing into commercial fabrication of a whole line of pressure-measuring equipment for potential clients outside the Pakistan nuclear program. If he went that route, he would have to make better equipment and demonstrate its reliability. He went for the first option because Kahuta desperately needed those transducers. The items used at Urenco and USEC (the US-origin originals) have a track record of reliability and performance. KRL’s reverse-engineered gear were cheap knock-offs. The transducers were less reliable, and a lot of them crapped out after short lifetimes. The commercial venture in transducers wasn’t a big success…He didn’t overtake the Western firms in the market. Instead he sold less reliable gear to customers with less money. They weren’t necessary buying for nefarious purposes so export controls weren’t an incentive for them to buy copycat goods from Khan. The reverse engineering dilemma can also imply the following choice: Do I make third-rate equipment that I can use now which will break down and have to be replaced soon, or do I instead forgo that problem and find an inventive way to continue to import reliable equipment which is (like the transducers) on NSG country commodity control lists? That’s a separate issue from the one Khan faced in deciding whether Kahuta or his perceived commercial market was paramount. It is also related to the issue whether to go for crash production of enough fissile material in the short term or instead tailor my R&D effort to result in a more sustainable outcome.
Because of the numerous threads at ACW that mention our theory, we finally decided to add some comments.
First of all, thank you very much for the interest in our ideas. It is very important to understand that we do not want to insist on our opinion just for our ego. Rather, we want to find the truth behind the whole story to give a realistic threat assessment of the countries of interest. We do not want to blame certain institutions – our analysis is based on technology only and has nothing to do with politics. And this analysis clearly points into certain directions.
Geoff Forden is basically right in his summary, but a sixth point should be added:
6) The various DPRK missiles differ totally from each other. Successful “developments” are abolished and re-invented from scratch. There is no clear line of development visible in the DPRK missile programs.
To clarify this point with two examples: If the DPRK had the Scud-engine available in large quantities, Nodong should have been outfitted with a cluster of 4 engines. Instead, they develop a new engine that is too weak for significant performance increase. Or TP1: They abolish the whole concept and develop a completely new SLV from scratch.
Countless missiles were developed in the Soviet Union of the 1950s and 60s. Many never reached the stage to obtain an R-XX number. Some got their number but were canceled after several tests, others were severely modified and still retained their number. We know, for example, of an early R-17 with engine parameters that differ significantly from the later 9D20. The missile was already designated R-17, though.
We all know that the Soviet missile history of this period is full of unknown unknowns.
It would be very helpful if one of the Soviet missile engineers of old would break the silence and comment our theory, even if it was anonymous.
If the DPRK had the Scud-engine available in large quantities, Nodong should have been outfitted with a cluster of 4 engines. Instead, they develop a new engine that is too weak for significant performance increase.
The scud-b engine is 130KN whereas the NoDong-a is a 280KN engine. It seems obvious that the DPRK felt a four S5.2 cluster would result in a vehicle without a mission. Too small to be a microsat launcher, not enough range to strike CONUS and too non-mobile and expensive with too many failure modes(4) to field as a long range MRBM. A lose-lose propisition. It seems to make far more sense to create a familiy of missiles based on the NoDong-A; a decent and cheap single engine MRBM and a four cluster ICBM.
As far as the potentional of the scud S5.2, it’s the Soviets that made the first pass at that; they never choose to make a four cluster of that engine, deeming it lacking in terms of growth avenues, how can we fault the DPRK for arriving at the same conclusion?
And with respect to the TD-1, what TD-1? It was a one off space shot that failed. THe TD-2, i.e. the Unha-2, was also a two off space shot that failed. Neither are production models but rather long assmebly time propaganda weapons, which have suffered a serious case of blow back. Both the TD-1 and the TD-2/Unha-2 launchers use the NoDong-a engines. I don’t see the discontiuity in the DPRK program, I do see a cash starved nation, crawling along with the occasional surge when geopolitics and economics align, using military rockets to score a space coup and failing badly.
And just to illustrate the connection, it has been argued here, not by me, that the Safir-2 is the culmination of work on the TD-1; an improved upper stage and a highly tweaked first stage.