Andreas Persbo has a “good post”:http://verificationthoughts.blogspot.com/2007/03/entering-kahns-little-den.html up at his place on the subject.
This is a decent summary:
bq. But what is the capability of this plant today? Well, figures are both unreliable and, in addition, some sixteen years out of date. It is *sometimes assumed in the literature that the facility has some 3,000 centrifuges operating at any given moment, and that the capacity of the plant is somewhere between 9,000 and 15,000 separative work units.* These figures rely on open source information compiled and assessed by the Stockholm International Peace Research Institute. (Albright, et al., 1997). The authors of the study readily admit that “the number of operating machines at any one time is highly uncertain. Part of the confusion is that Pakistan has installed considerably more machines than it has successfully operated. In 1986, it was reported that Kahuta had 14,000 centrifuges.” (See p. 275). Yet, *the claim that the plant is capable of a meager 15,000 SWU’s resurface time and time again.* Over the last couple of weeks, I have been compiling a Google Earth placemark of the A.Q. Khan Research Laboratory, cross-checking with public imagery analysis to get a complete picture of the facility. Almost immediately, *I suspected that the 15,000 SWU figure might be grossly misleading.*
The argument is a little involved, so read the whole thing. I was glad to see that he gave proper credit to “Jeffrey and Mark Hibbs.”:http://www.armscontrolwonk.com/1399/p3-and-p4-centrifuge-data