Jeffrey LewisIran's 164 Centrifuge Cascade

Somehow this news story didn’t get wider attention, but Dafna Linzer’s article in the Post last week claimed some of Iran’s centrifuges appeared to crash during the operation of the 164 centrifuge cascade.

Linzer cited “several officials with knowledge of the nuclear program said yesterday that the cascade was no longer operating and that a number of the networked centrifuges had crashed during a fairly rushed process.” [Emphasis added]

Aqazadeh mentioned some problems in the process, although he gave no reason to think the issues will arise again:

I’ll give you an example. In the preliminary stages of the work, we noticed that our machines broke down frequently. We couldnʹt discover the cause, since we didnʹt have any scientific sources or books to refer to. After great efforts we discovered that our experts didnʹt wear fabric gloves during the assembly phase. We found out that when you assemble the parts with bare hands, germs are transferred to the machinery from the smallest amount of sweat which comes off the hands.

This little amount of germs is enough to trouble and destroy the machine. When we say a machine is destroyed we mean that it turns into powder.

I would be fascinated to find out what else has been tough for the Iranians.


  1. John Field (History)

    This is fishy.

    Good vacuum practice always indicates the use of gloves, careful cleaning procedures, and usually an extended bakeout under vacuum after assembly. This has nothing special to do with uranium centrifuges.

    Fingerprints are loaded with water, hydrocarbons, and salts. UF6 probably reacts with all of these things and HF is one likely product. And, 7000-series aluminum alloys have poor corrosion resistance. Super strength scandium-aluminum alloys might be much better in this regard.

    I guess it is possible that fingerprints could accelerate corrosion at the spot which would eventually weaken the rotor wall at the fingerprint.

    Note that this problem would not occur when testing the machine under vacuum.

    On the other hand, this is very sloppy, and at that level, there are many things which seemingly might have a greater impact – like lubricating oil composition, cleaning solvents, welding and coatings, seal materials, and purity of the UF6 itself(as Jeffrey mentioned before)

    Furthermore, once you build a centrifuge, wouldn’t you test a sealed one in UF6 for a while to check for aging processes?

  2. Mike

    Good that you ask! I stumbled upon that Linzer line too; but somehow I interpreted it (probably wrongly) to be a reference to the old IAEO report or an old David Albright quote. There is no reference to any broken centrifuges in the very last IAEO report. Seems to me, that this information can come only out of Vienna. So, maybe someone can ask Linzer directly, where he’s got that from….

  3. Ali

    Iran has confirmed the crash but says the centrifuges affected were repaired. I don’t know where I read this; it was a Persian language web site/news source.

  4. Russell (History)

    One of the preparatory processes that is required before using a centrifuge component for the first time is “passivation” – which basically involves bathing any UF6 exposed bits in UF6 so that anything with a remaining potential to react will react in a controllable environment rather than in the vacuum system. The presence of fingerprints would be enough to interfere with the passivation process.

    Strangely enough the presence of fingerprints would also affect the balance of the rotor directly – it may not be obvious to most people but fingerprints weigh something – at 350m/s that can be a surprising amount.

    Agazadeh’s description is of old problems (before the suspension) – even without the problems that he discussed (such as no gloves), getting even P1 centrifuges to balance is difficult. If their crash rate on start up is 1 in 10 or less they are doing remarkably well.

  5. John Field (History)

    As Russell mentioned above, I wondered also about direct imbalance effects, but I reasoned that it is too small as follows:

    Fingerprint thickness must be in the range of hundreds of angstroms.(otherwise my fingers would just erode away tapping on a clean dry surface!)

    Specific gravity could at most rise to something approaching UO2 due to cracking of UF6 on the surface.

    So, we’re talking something like 10’s of micrograms mass. That translates into tipping the rotor primary axis by maybe a few hundred angstroms.

    I suppose also there would be some acoustic power radiated into the UF6. Not much at only 1 Khz though.(there is the irritating possibility of acoustic resonance in the UF6, but presumably you’ve got to avoid that anyway.)

    It just doesn’t seem like any of this is enough to be a big deal. Besides, the whole business of slathering gooey fingerprints over all this delicate vacuum machinery is sloppy as hell – to the point of being technically ridiculous.

  6. Yale Simkin (History)

    This whole process is simply learning curve.
    Thats why you start with a prototype, then a pilot, then production.

    Just like adding straightforward processing procedures to remove excess molybdenum from your feedstock, installing and operating your cascades properly should be a quick job.

    Minor speedbumps like this will be irrelevant a few months from now.

    As Supreme National Security Council secretary Hassan Rohani pointed out:


    “What makes the nuclear discussion sensitive is that the nuclear technologies and technologies related to the [nuclear] fuel cycle are particularly sensitive technologies.
    This is because having fuel cycle capability almost means that the country that possesses
    this capability is able to produce nuclear weapons, should that country have the political
    will to do so. This means that a country that possesses fuel cycle technology can enrich uranium, and the country that can enrich uranium to about 3.5 percent will also have the capability to enrich it to about 90 percent. It is for this reason that today the international
    circles are discussing this capability with an extraordinary degree of sensitivity.
    They would view a country that possesses this capability with concern and a sensitive eye. The usual practice is to put pressure on a country that is standing on the threshold of this technology. That is to say, if a country does in fact fully develop this technology, it would be very difficult, if not impossible, to continue the pressure.”


    We are smackdab at the exact moment when “it would be very difficult, if not impossible, to continue the pressure.”

    (A fact of which the Israelis are acutely aware)


  7. Russell (History)

    Note for John Field,

    Have you ever used a spoke key to try to “tune” (balance) a bicycle wheel? As you get closer to perfection you start to find that ridiculously small adjustments make noticable differences to the balance and we are only talking rotational speeds of the order of 10m/s.

    As I noted earlier people don’t seem to realise what sort of mass change (acceleration) we are talking about here. If the diameter of the rotor is 100mm (or 105mm if we use Agazadeh’s figures) and the rotational speed is 350m/s the formula for the g-force is
    (v^2/r)/9.8 =

    ~ 125,000 g

    To put this in perspective a fingerprint that consisted of 0.00001g (your 10s of micogram level) of oil or sweat would end up weighing of the order of grams at 350m/s on a 100mm rotor (an imbalance comparable in effect to the valve on a racing bike tire at 10m/s). Any fingerprint or sweat droplet that was actually observable would weigh in at the 10gram or 100gram level.

    Any mass imbalance becomes important as you aim for perfection – especially for super-critical rotor designs.

  8. Otfried Nassauer (History)

    As to the source for the more recent reports about crashing centrifuges:

    Ali Akabar Dareini of AP reported on May 2 from Teheran:

    “Ghannadi said Iran’s enrichment of uranium was continuing, but he confirmed reports that a few of the centrifuges at the enrichment facility in Natanz had crashed last month.”

    “It’s not a problem. They were repaired,” Ghannadi said in this holy city south of Tehran.

  9. mark gubrud (History)

    This discussion reminded me that I was puzzled why centrifuges should be characterized in terms of rotor rim speed instead of acceleration, hence g-force; now I think I understand.

    The basic figure of merit, for comparing different designs from the point of view of separative work, should be the rim acceleration times the rim area (length times pi times diameter).

    But length is less important if we assume that the bearings, plumbing, drive and supports take up a minor fraction of the total length, so that shorter units can be stacked on end.

    Then the basic figure of merit, which characterizes the level of technology, is acceleration times diameter, the square of the rotor rim velocity.

    Thus a unit with ten times the velocity is 100 times more effective, etc.

  10. Russell (History)

    Note for mark gubrud,

    Actually the efficiency of a centrifuge goes up by the fourth power of the rotational speed – ie. 10 times the speed gives you 10,000 times the effectiveness.

    The difference between 300m/s and 350m/s is a factor just a bit less than twice the efficiency.

  11. mark gubrud (History)

    Several people have written to me pointing out that the separative power of a centrifuge is actually proportional (at least roughly) to the fourth power of rim speed, rather than the square. But I haven’t seen a simple argument to explain this.