Further Comments Regarding the BAS Article on Fordow
by David Albright and Paul Brannan
December 4, 2009
We wanted to share some thoughts on Oelrich and Barzashka’s justification for their analysis in the Fordow article released November 24 in the Bulletin of Atomic Scientists. In this justification which they published on the Federation of American Scientists (FAS) web site on December 1, they assert that the best estimate of the near-term capacity of the Fordow plant is the most recent capacity of the Natanz facility. However, this assumption is almost guaranteed to underestimate the actual capacity of the Fordow plant. It ignores the effect of what Iran has learned through the operation of the centrifuge cascades at Natanz in terms of improvements in cascade performance and centrifuge construction and installation. This analysis also does not take into account that Iran would expect a higher level of performance out of the IR-1 centrifuge by 2011.
But a more fundamental problem in their analysis is that they significantly underestimate the performance of the Natanz facility. Applying their estimate of the enrichment output at Natanz (given in the FAS analysis) results in a significant underestimation of current production of low enriched uranium (LEU) at Natanz. Taking their 0.44 separative work unit (swu)/year per machine and applying it to the number of centrifuges stated by the IAEA to be enriching as of November 2, 2009, namely 3,936, the total enrichment output, would be 1,732 swu/year. The authors also assert that the tails assay at Fordow should be 0.25 percent, a highly questionable assumption based on Iran’s experience at Natanz. Nonetheless, applying this tails assay and taking the enrichment level of the product as 3.5 percent, the plant would then be projected to produce 532 kg LEU of hexafluoride a year. For the last nine months, however, Natanz has produced on average about 83 kilograms of LEU hexafluoride per month, which on an annualized basis corresponds to approximately 996 kg of LEU hexafluoride, almost double their estimate. Applying the tails assay used at Natanz, about 0.4 percent (based on Iranian declarations and supported by environmental sampling), gives a projected value of 704 kg of LEU hexafluoride, a value still significantly below actual LEU production.
We were unable to understand the problems in the FAS calculation. But one problem is that the data contained in IAEA reporting is insufficient to allow accurate estimates of the average enrichment output of the centrifuges using the methods pursued by Oelrich and Barzashka. The data’s incompleteness is the main reason we have avoided the type of calculation that FAS performed. For example, calculating the mass balance on the uranium 235 (uranium 235 in the feed should equal the uranium 235 in the product and tails) is not possible based on the available information. This requires assigning values in a formula that are impossible to substantiate. FAS appears to have forced a U-235 mass balance by adjusting the tails assay in Table 2 in their assessment to 0.46 percent as a way to get the masses to match. But the situation at Natanz is quite complex. Without more data, this type of calculation is prone to error and misinterpretation. Another difficulty is determining the number of centrifuges enriching during extended periods. The IAEA publishes the number of centrifuges that Iran says are enriching on a few of the days during the year. However, these numbers are only rough estimates of the actual number of centrifuges enriching on a sustained basis. This aspect is further discussed here.
The authors also appear to have ignored the impact of the tails assay estimate in Table 2 when they later decided to use 0.25 percent in estimating weapon-grade uranium output at the Fordow plant. In that case, they used a tails assay of 0.25 percent without any discussion of why they ignored their earlier estimate of the tails assay. This choice, along with the decision to use 0.46 percent tails assay in the earlier part of their assessment, has the effect of further lengthening the time needed to produce weapon-grade uranium at Fordow. As such, the authors’ estimate is highly biased toward the far lower range of possibilities. Because of this bias, we believe that the authors’ assessment about Fordow is unrealistic.