Jeffrey LewisGuangyuan Plutonium Production Reactor

Google Earth now has high resolution imagery of what is generally believed to be China’s Guangyuan plutonium production reactor. Click on the image for a .kmz file.

The annotations are reproduced from David Albright and Corey Hinderstein, Chinese Military Plutonium and Highly Enriched Uranium Inventories (Institute for Science and International Security, June 30, 2005). David and Corey were the first people to publish commercial images of the facility.

David Wright and Lisbeth Gronlund also provide a nice summary of the reactor’s operating history in A History of China’s Plutonium Production (Union of Concerned Scientists, January 16, 2003).

Although China has never disclosed the amount of plutonium, classified DOE estimates leaked to the press suggest China produced 1.7 – 2.8 tons – consistent with the low end of open source estimates.

China’s plutonium production reactors may have suffered operating probelms evident to the US intelligence community. Both sets of authors agree that operating histories might result in lower estimates, as Albright and Hinderstein explain:

Total production is therefore estimated as 2-5 tonnes of weapon-grade plutonium. Because of the lack of hard evidence on the production facilities, in particular on the power and operating histories of the reactors, Wright and Gronlund state that their estimates have a high degree of uncertainty.

Wright and Gronlund specifically mention rumors of a “a fire during the 1970s that seriously crippled” one of the reactors.

Declassified U.S. intelligence documents confirm that China’s plutonium production facilities encountered significant technical problems. The declassified report, China: Plutonium Production Reactor Problems (CIA: January 1988), is almost entirely redacted … but the title kind of sez it all, don’t it?


  1. David Albright (History)

    Hello Jeffrey

    I think what you have written contains several inaccuracies about our work and others. Our estimate of Chinese weapon-grade plutonium production is 2.3-3.2 tonnes, close to reported DOE estimate of 1.7-2.8 tonnes. In addition, we do not know if the DOE estimate was reported correctly; other numbers for different countries leaked in the same media report were wrong. With regards to operational problems in the reactors, that issue was central to both Wright and Gronlund’s and our analysis and is clearly mentioned in our reports. Your comment about South Africa’s production of HEU in the 1980s is likewise misleading.

  2. Jeffrey Lewis

    Per David Albright’s request, I am posting an e-mail exchange with him:

    ——- Forwarded message ——-
    From: “David Albright”
    To: “Jeffrey Lewis”
    Cc: “‘Jacqueline Shire'”
    Subject: RE: Updated post.
    Date: Sun, 25 Apr 2004 08:22:48 -0400


    You did not include our estimate of weapon-grade plutonium production. You merely quote part of what we wrote about David and Lisbeth’s study. Our estimate is further back in the paper, where the wgp estimate is as I wrote yesterday, 2.3-3.2 tonnes. We also estimated the possible production of lesser quality plutonium after China’s halt to the production of plutonium for weapons.

    Surprisingly, you did not include my comment on the data from Gertz. As a matter of practice, information from Gertz should be approached with skepticism. For example, in Gertz’s table, included with the BAS article you linked to, the number listed for British wgp is wrong. His table lists under the wgp column all declared British military plutonium, including non-weapon-grade plutonium. The DOE certainly was aware of that, particularly since those numbers were made public in 1998. In addition, much of the North Korean plutonium was non-weapon grade, and this fact was well known to the DOE. I believe the table is a mish-mash of information, some reliable, some not. For example, the US value for wgp is accurate. But for
    states with secret programs, such as China, it is hard to know whether or not the Gertz value is reliable. Thus, I do not think you are justified in attributing such uncritical support to the accuracy of Gertz’s value, particularly as a pivot to make judgments about the accuracy of our work.

    In the end, I find your reposting misleading. As a result, please post my comment yesterday and the one above as I wrote it.


    —– Original Message —–
    From: “Jeffrey Lewis”
    To: “David Albright”
    Cc: ‘Jacqueline Shire'”, “David Wright”
    Subject: Comment
    Date: Mon, 13 Nov 2006 12:17:07 -0500


    I am happy to post your comment, but perhaps you’d prefer me to change the wording of the post instead.

    If I have offended you in some way, I’d rather resolve the issue in private rather than on the website. If you have a complaint about something I write, you are always welcome to send me a note or call me (my cell phone is XXX-XXX-XXXX).

    I am also cc’ing David Wright, to give him the same opportunity

    My post is not a criticism of your estimates, which I believe are correct given the available evidence. I only wanted to observe for readers the caveat that appears in your work and the Wright/Gronlund article, that operating problems could reduce estimates and note the available evidence for that possibility.

    Let me try to be somewhat more transparent about my thinking — and feel free to tell me where I am being inaccurate. Open source estimates of Chinese production range from about 2-5 tons, with the Crystral Ball Method producing a somewhat smaller range:

    Wright/Gronlund 2-5 tons Albright/Hinderstein 2.3-3.2 tons (95 percent confidence interval).

    I suppose my thinking was also somewhat affected by your pre-Crystal Ball estimates (Albright/Kramer: 4.8 +/- 2 tons; Albright/Berkout/Walker: 2-6 tons)

    Thus, it seemed perfectly innocuous to me to describe the DOE number of 1.7-2.8 t as falling “at the low end” of the two papers I cited and the open source estimates generally.

    As I say, I will reword the post to accomodate a correction of your suggestion. One option is for me to quote one of the articles mentioning the potential impact of operating histories as a transition in order to make clear to readers that you discuss this possibility.


    As for South Africa, I am going back through my notes to see where I got the idea the idea that US IC HEU estimates for Valindaba were high for the 1980s.

    I seem to remember the US IC was not convinced about the completeness of the SA declaration, which suggested their internal numbers were higher. Now, where I got the idea that operating history was important, I will try to find out …


    I still think I am right about the South African estimates, by the way, although I will save that for a later post.