Estimating How Many Pits DPRK Has, Part 1.5 New Numbers

Jul 30, 2018

In Part 1 I had a guesstimate at the production capacity of North Korea to make fissile material and other relevant materials for their nuclear weapons. I wanted to talk a little about their testing regime and also about some considerations about bomb design, missiles and missions, and I've been working on that, but some things have come to light that I want to discuss first.

First and foremost, I missed a really, really important pdf from David Albright at ISIS on Kangson and it's production capacity based on his figures and on US Government figures that's he's privy to. And secondly, I'm not happy with my HEU numbers and I think I made a mistake. I don't think it's right to stealth edit out my figures so I'm going to keep them and cover Albtight's figures as well as my new production numbers.

Albright's Numbers

As I said above, I think Albright has a very interesting pdf filled with new data. Personally if anyone is close to knowing their stocks of fissile material, it's Albright so with that in mind, I want to look into what he has, discuss it and take it into account for further discussions on estimating how many pits North Korea is capable of producing.


Albright accepts, though with some skepticism, the numbers of his US Government sources, giving an estimate of "6000-12000 or more P2 centrifuges" which I think is A LOT and given that he thinks it's capable of 600Kg to 1000Kg of HEU production, I'm way off. Even my wrong numbers in my last post don't come cose to that. Even with my fixed estimates I'd come more around 240Kg to 480Kg.

Now Albright's a much smarter guy than me, and I'm more inclined to believe his numbers rather than mine, but I'm having trouble squaring this circle since I can't verify his numbers. Is he talking in terms of cascade's or centrifuges? Or interchangeably using cascade and centrifuge? Because if Kangson is packed with a cascade per square meter rather than a centrifuge per sqm, then all of a sudden, you have a lot more centrifuges and a lot more production bringing those numbers into closer to my current new estimates below.


Albright has a hard cap on 30Kg of Pu processed from unloaded fuel at the 5MWe reactor. I'm inclined to believe this as Chris mentioned that Reprocessing is a dirty, nasty, and very visible process and the reprocessing facility at Yongbyon has been quiet in imagery and pretty cold in thermal signatures implying that there is very little going on in terms of reprocessing. Also interesting to note that Albright doesn't think that much has gone on in the way of reprocessing since 2007 when he estimated that North Korea had 28Kg to 50Kg of Pu 239 that was usable.

Other Notes

One of my goals with this series of posts was to either verify the DIA's estimate of having 60 pits but more importantly, that North Korea was making enough fissile material for 12 more each year. The report "assumes the use of composite pit core designs". Albright had some thoughts on this;

In one report, the U.S. indicated that North Korea had up to 60 nuclear weapons. In our analysis, I would interpret this value as not including losses and being in the upper tail of the first distribution.
I would stress that in our analysis a value of 60 represents a worst case.
And I would also stress that our base estimate is 14-34 nuclear weapons, reflecting on-going uncertainties about the status and operation of an older centrifuge plant.

This would be where my thinking is going. It doesn't seem that North Korea has that kind of production capacity. Albright says that it's an absolute worst case, which may imply a third site, which is the other goal of this series. If I can't find a way to verify the DIA's estimate, then that's because they know more than they are letting on about North Korea's capacity which opens up routes for further analysis on suspect sites at locations like Hagap, Bakcheon, or Taecheon to name just a few as the NTI database has many suspected nuclear sites with unknown uses from reactors to enrichment facilities.

Correcting My Numbers

As much of a valiant effort as I made, in making my numbers, I believe that during the first drafts of of the first post, I had estimated a LOT higher numbers of centrifuges for Kangson's capacity, and maybe I was right given what Albright and the US Gov estimated, but I never changed my estimates after rounding down the number of centrifuges. So I want to correct that. I also think that unless the P-2 centrifuge is between 0.5x and 0.75x smaller than the P-1 it's highly unlikely that at 1 square metre per centrifuge, that there are 6000 to 120000 centrifuges in that building. But as I have said it's unlikely that I'm right and Albright is wrong.

The long and short of it is that Sig Hecker say's that Yongbyon is capable of making 40Kg of 90% HEU with it's 2000 centrifuges. That's 40 grams per centrifuge. If that's the case, then the total amount of capacity from my estimate of 6675 centrifuges is 267Kg total and for Kangson with 4475 centrifuges, that's 187Kg. As I also said above, if cascade and centrifuge are being used interchangeably, then there's reason to believe this is the case, but then we're looking at 3 or 4 centrifuge's per sqm, which makes the numbers make sense, but that again requires that cascade and centrifuge are being used interchangeably.

Though if you think these numbers are straight up wrong, ping me with an @legendarypatman on twitter or on the ACWP Slack thread or contact me via the About page because as close as my original numbers seem, I don't think I can justify them.

New Totals for Yearly Production

Finally, an updated table of production. Pu is still a static number as the 5MWe reactor is still seems unused or has been running very minimally and the ELWR doesn't seem to be ready to run just yet.

Material My Minimum qty My Maximum qty Albright/USG Min Albright/USG Max
HEU 187Kg 267Kg 600Kg 1000Kg
Pu 56Kg - 77Kg 76Kg - 103Kg N/A 30Kg
2H ??? ??? N/A N/A
6Li 120Kg ??? N/A N/A
3H 60g (IRT-2000 only) ??? N/A N/A

This post wouldn't have been possible without a great conversation that Peter started in the ACWP Community Slack as well as help from Peter, Andrew, Retin, Chris, Nathan, Kieran and Dylan. Massive thanks to you guys for the entertaining conversation, facts, fact checking and finding awesome sources!

Paddy Kerley

DFCS at TU Dublin's Cyber Program and talk giver TU Dublin Hacker Soc. Mostly talks about Cyberwarfare, NatSec, Infosec, Big Safari, Arms Control, North Korea, OSINT and stuff that goes boom

Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.