Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
-TS-/
N orth Korea Centrifuge Capabilities (U )
Derived From: Multiple Sources Declassify On : August 3, 2034
St udy Leader -Contr ibutors Include:
!DOE (b)(3), (b)(6) I
JSR-09-510
October 21, 2009
!DOE (b)(3) I
TSf~/NOFORN~
jDOE (b)(3) I
85030-007 /09 Total Pgs: 124
Cy ___ of __ _
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
IDOE (b)(3) I -'ffli ~/NOFOIU~ fltBQ3
Contents
85030-007 /09 Page iii
1 (U) EXECUTIVE SUMMARY 1
2 (U) INTRODUCTION 6 2.1 (U) JASON Charge . _ .. _ . . 7 2.2 (U) Organization of this Report 8 2.3 (U) Acknowledgments . . . . . 9
3 (U) MANUFACTURING APPROACHES 10 3.1 (U) Modern Manufacturing in the Developed World 10 3.2 (U) Avoiding Mirror 1maging . . . . . . . . 11 3.3 (U) Best Guesses for NK Practices . . . . . . 13
3.3.l A previous US effort at avoiding mirror imaging 15
4 (U) CENTRIFUGE BASICS 16 4.1 (U) Basic Ideas Due to Zippe . .. .. _ 16 •L 2 ( U) Physics Performance and Scaling Laws . 17 4.3 (U) Importance of Rotor Material Strength . 19 4.4 (U) Maintaining the Counter-Current Flow 20 4.5 (U) Cascade Priuciples . . . . . . . . . . 21 4.6 (U) Instabilities and Bellows . . _ . . _ . 22
5 (U) WHAT ARE THE LONG POLES? 25 5.1 (U) Precision or exotic machining _ . . . 25
5.1.1 (U) Stationary magnet holder clips 28 5.1.2 (U) Spiral grooves on the ball bearing 29
5.2 (U) Frequency Converters and Motors . . 30 5.2. l (U) Commodity audio amplifier chips 32 5.2.2 (U) Commodity microcontrollers 32 5.2.3 (U) Commodity power supplies 33 5.2.4 (U) 0!.her features . . . . . 33
5 .3 (U) Rot.or Material and Fabrication 34 5.:3 .l (U) Maraging steel . . . 3-! 5.3.2 (U) Flow forming . . _ . 36 5.3 .3 (U) Final hydroforming . 37
5.4 (U) Bellows . .. _ . . . . . . . 38 5.5 (U) Centrifuge-Specific Know-How 39
"PS-/ - /NOFORN ftZRQJ !DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
!DOE (b)(3) I 'TS-( ~/NOFORN ftt8Q)
5.5.1 (U) Upper Suspension Mag.nets . 5.5 .2 (U) Rotor testing and balancing . 5.5.3 (U) Bearing and damping testing 5.5.4 (U) Secondary flow and testing
5.6 (U) Manufacturing Scale-Up . . . 5.7 (U) Human Factors .. . . . .. . 5.8 (U) The North Korean Economy
6 (U ) ESTIMATED TIMELINES 6. 1 (U) How Long Does it Take? . 6.2 (U) W here Are They Now?
7 (U ) INDICATORS 7. 1 (U) Making Centrifuges 7.2 (U) Operating the Centrifuges .
(U) Appendix A: Briefing Slides and Figures
(U ) Appendix B: CHOP
85030-007 /09 Page iv
39 40 42 43 46 47 ~8
51 51 53
54 54 56
58
110
(U ) Appendix C: Towards Better Estimates of Program Size 113
(U ) Appendix D: Size of an Enrichment Cascade Facility 115
(U) Appendix E: Motor Windings 116
-T&f ~/NOFOR:N lf±BQJ IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -TS"f~/NOFORN ~
1 (U) EXECUTIVE SUMMARY
~~/NOFOR1'ilt!:8Q] jDOE (b)(3) I
85030-007 /09 Pagel
lDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I
-1!fSj ~//NOFORNltiBQ)
-~ / NOFORPJ' fltBQ]
!DOE (b)(3) I
85030-007 /09 Pttge 2
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
-
85030-007 /09 Page 3
100E (b)(1) I
· ~ •
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -TSf / NOPORN" lftBQ]
~/ ~/NOPORN fllB:Q] jDOE (b)(3) I
85030-007 /09 Page 4
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I .!:£5 / / MOFORN ~ 85030-007 /09
Page 5
(U / / FOUO} Our report concludes with a. list of possible iJ1telligence
indicators for the various steps, most already well-known; these are not re
peated m this summary.
"'8( ~(/NOPORN ~ DOE (b)(3)
lDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!£5/ 85030-007 /09 Page 6
2 (U) INTRODUCTION
(U //FOUO) On Juoe 15, 2009i responding to threatened UN sanctions,
North Korea stated, "enough success has been made in developing uramum
enrichment technology .. . to allow the !necessary?] experimental procedure.
The process of uranium enrichment will be commenced."
:ff!/ ,,_,f NOFORrtlii!fil]
100E (b)(3) I
100E (b)(1) I
IDOE (b)(1), (b)(3) I
IDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
jDOE (b)(3) I --TS-/ ~/NOFOR.N ~ 85030..007 /09
Page 7
2.1 (U) JASON Charge
(U //POUO) The formal charge to JASON is as follows:
(U//FOUO) "The Director of Nalional Intelligence's Nort.h Korea Mis
::;ion Manager. in conjunction \lr'1th the Nat.ional Counter Proliferation Center
and the National J ntelligence Officer for Wea.pons of Mass Destruction re
quest the JASONs undertake a summer study on the t.opic of Nort.h Korea ·s
technical prowess and ability to construct and operate a uranium enrichment
centnfuge capability. The Study will have as goals:
• (U //FOUO) Determine if Norr,h Korea is technically capable of build
ing centrifuges, constructing an enrichment faci1ity. and operating ce11-
trifuges with little or no outside assistance.
• ( U //FO UO) Examine those North Korean industries, processes. and/or
technologies that are analogous or might. provide a useful fouudation
to those needed in centrifuge con~truction and operation."
-rpfj-f / / NOFORN ~ !DOE (b)(3) I
100E (b)(1) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
::rs;
2.2 (U) Organization of this Report
85030-007 /09 Page 8
{U//FOUO)The PowerPomt briefing slides that accompany this report
as Appendix A are an integral part of the report. While the narrative text in
this report is self-contained, we will not repeat in the report body the figures
and photographs that are in the briefing slides; rather, as they are discussed
they will be referred t.o here by slide number .
This report is organized as follows:
(U//FOUO) In Section 3, we set the stage by discussing die large differ
ences between modern (say, post-1985) manufacturing practices in the indus
trialized world . a.nd those practices that are likely to be the uorm in North
Korea, a. count.ry whose industry is mostly (but not completely!) caught in
what we might think of as a pre-1960 time warp. IODNI (b)(3) I
(U//FOUO) In Section 4, we rev iew the basics of gas centrifuge physics
and engineering. \Vhile this topic is covered in many open-source publica
tions, the summary given here is necessary for understanding the basis of our
subsequent con cl us ions.
(U //FOUO) In Section 5 we discuss m turn the various "long poles'' or
hurdles that may fa.cc NK, and assess whether and how they might reasonably
be expected ro overcome them .
(U //F'OUO) In Section 6, we attempt to consolidate our assessments
into a notional tinieline 1md to draw some concl usions.
~ / /NOPORN lft8Q] IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~/ ~/NOFOftN lftB:Q] 85030-007 /09
Page 9
(U //FOUO) In Section 7, we suggest intelligence indicators (mof.>t al
ready recognized by others) that could potentially shed light on where N K
is along tt c timeline.
2 .3 (U) Acknowledgments
~/ ,,_./NOFORNl!tffiB DOE (b)(3)
100E (b)(6) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~ ~//NOPORN lft8Q] 850.30-007 /09
Page 10
3 (U) MANUFACTURING APPROACHES
(U//FOUO) We begin ·with a discussion that might seem peripheral
r.o the subject at hand. but. , we think, is in fact central: How does modern,
U.S. and developed world. manufacturing practice d iffer from what we expect
North Korea to be capable of?
3.1 (U) Modern Manufacturing in the Developed World
--!:f&f ~/NOFORN~ IDOE (b)(3) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -'fflj ~/:NOFORN ~
-
3.2 (U ) Avoiding Mirror Imaging
re;
85030-007 /09 Page 11
IDOE (b)(3) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I :fS/ / / NOFORNftfilQJ
'P'!i / /M OFOilJ!'tlti:BQ)
IDOE (b)(3) I
85030-007 /09 Page 12
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
jDOE (b)(3) I -'¥3-/ /NOFORN~ 85030-007 /09
Page 13
3.3 (U) B est Guesses for N K P ractices
(U //FOUe) We have attempted to crystalize t.he discussions of the t.wo
preceding sections into 11 set of plausible, perhaps even likely, design and
m<tnufaccuring practices that Nort.h Korea might bring to bear on a gas
centrifuge effort. T hese are the practices that, in the rest of this report , we
\Yill apply to the specifics of centrifuge de.sign .
• -'PS"f~/NOl'OJtN l@QJ
DOE (b)(3)
100E (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
!DOE (b)(3) I "£.&/ / NOFOR:N ltt:BQ]
-
!DOE (b)(3) I -rps.; ~/NOFORN~
85030-007 /09 Page 14
!DOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
IDOE (b)(3) I ~ ~/t<itOFORN!tlfilE
3.3. l A previous US effort .at avoiding mirror imaging
.::f-8 I /.NOFOR.?~
85030-007 /09 Page 15
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I 4:Sj ~/toqOFORN·~ 85030-007 /09
Page 16
4 (U ) CENTRIFUGE BASICS
(U//FOUO~ We now tum to the phvsie:s and eng,ineenng basic:-; of gas
centnfugc design . The figures in Slides 10 and 11 should be consulted at this
pomt.
4.1 (U) Basic Ideas Due to Zippe
(U//FOUO) The enduring basic ideas of uranium separating gas cen
t.rifugcs arc due to Gernot Zippe, who dcw~J oped them in the USSR in the
1940s <ind 1950s. and ~hen re-implemented them in the US in the late 19.SOs
(U//FOUO) A Zippe-derived centrifuge has these key features (see Slides)
• (U//FOUO) It uses high g-forces to create a vacuum on-axis, where it
is fed . Thus seals between the rotati1tg <lnd fixed components arc:' not
needed .
• (U//POUO) It u::;cs a slow counter-current, circulating between r.hc
top <ind bottom of the vertical centrifuge to create what amounts to an
internal cascade (further explanation later), and to obviate the need
for precise positioning of the prod uct. and tailing scoops in t he radial
dirPct1on .
...r.r&f~/NOFORNfltBQ] jDOE (b)(3) I
IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
!DOE (b)(3) I -I:PSf / NOFOR:N fH:BQJ 85030-007 /09
Page 17
• (U/f FOU 0 ) Its top magnetically suspended bearing na.t.urnlly allows
acceptable wobbie wit.hont damage , 11nd also reduces the load on the
bottom needle-bearing.
• (U //FOUO) ft maintains vacuum ruound the cylindricaJ rotor by a
simple helical groove. the so-called "molecular pump" 1 machined into
the \'acuum case . The molecular pump has no moving parts.
• (U /lf'OUO) It is powered by an electric motor that uses the bottom ro
t.or end-cap as a passive magnetic rotor: there are no rotating electrical
components.
• (U//fOUOj It may, but need not, have bellows that allow operation
above the first one or two resonance frequencies (more on this below).
4.2 (U) Physics Performance and Scaling Laws
(U//F OUO) Zippe-derived centrifuges involve an elegant interplay be-
tween their physic;:; and engineering features . The essential physics has two
parts:
(U//FOUO) First, under the high g-force of the rotation 1 heavier 1.1238,
as UF 0 , diffuses radially outward , while lighter U235 diffuses radially inward .
In equilibrium, t.his wouid produce a concentration difference only....., 1 % over
a useful densicy range .
( U//FOUO) Howe\·er , second, t.he counter-current causes enriched prod
uct to move slowly a..xia.lly upward, while deplet.ed ta.il ing-s move slowly axi
ally dow?1ward. Since the,..._, 1 % grndient hi refreshed radially during this s]o,v
drift. the result is a kind of cascade that produces a concentration difference
of as much as ,.,_, 20% between the top and bottom of the rot.or .
.!ffr] ~/NOF0ftl1~ IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -4:Sf ~//NOFOR1~ fllB:Q] 85030-007 /09
Page 18
(U//F OUO) The maximum rate of separative work of a countercurrent.
ga;3 centrifuge is given by Dirac 14]: 2
oU = D ((M2 - Mi)(wr)2) ~z rnar P "k 7. ?
~ ·B w
( 4-1)
where pis the gas density and D its self-d1ffusiv1ty (the product is indepen
dent of density for dilute gases), A12 and Af1 are the molecular masses of the
gases to be separated, u-' is the angular velocity, r is the cyllnder outer radius,
u = ..vr its peripheral speed and Z :s 1t.s length.
(U //FOUO) The separative power is essentially independent of the gas
density. althongh this ent.ers indirectly tbrough the effect of kinematic v1s
cosit.y (ex: l/ p) on the countercurrent flow pattern.
(U//FOUO} Expressed iu kg-SWU units (where 5000 kg-SWUs pro
duces about 20 kg of HEU) , equation (4-1) can be ""Tit.ten as
kg-SWU (300°I<) (J\!2 - M1 \ 2
( v ) 1
( Z ) ~10 -- ----) - . year T 3 amu 400 en/sec 1 m
(4-2)
In equation
(-1-2) one sees how t.he capacity of a centrifuge scales against nominal values
of length .z and rot.or circumferential velocity u, the two most important
parameters.
(U //FOUO) Empirically, according to rl.ata presented by Whitley [4).
pD ex T, al though elementary kinetic theory for monatomic gases would
suggest a scaling e< T 1t 2. The origin of the discrepancy may lie in the cou
pling between vibrational and translational degrees of freedom. In diatomic
molecules vibrational frequencies of bonds involving first period elements
such as N, 0 and F are generally at energies O(l000-2000) 0 1< x k8 , and are
not rnucb excited at room temperatures. In UF6 the frequencies ar lower ,
part.ly because with a heavy atom a one end of the bond the reduced ma.<:s is
greater by a factor of aLout 2, and partly because with a total of :-;even atoms
""'l'Sf I fNOPOitN~ IDOE (b)(3) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -rs-/~ / NOFOftl<: rtt:8QJ 85030-007 /09
Page 19
some of the modes a.re at lower frequencies than a simple diatomic stret.cbing
mode. Superelastic collisions may make the effective (momentum transfer)
mean free path an increasing function of temperature, thus increasing D at.
higher tempera tures.
(U//POUO) At fixed penpheral speed the separative power is indepen
dent of radius. This may appear surprising, but the relative drift speed is
proportional to the eentripet.11.l acceleration u2 fr, and hence inversely propor
tional to the radius, canceling tbe proportionality of the surface area. The
difference in equilibrium rni.x fractions depends only on l he peripheral veloc
ity and is independent of radius if the velocity is held constant . Cohen [3J
points out that the radius is chosen for mechanical reasons.
(U//POUO) We also not.e the dependence on temperature. This is not
steep, and is limited by the rapid falloff of UF6 vapor pressure as tt'mpera~ure
decreases (the equation is invalid and separatory power falls off rapidly if t.he
vapor density is so low that it t'nt.ers thr free molecular flow regime) . A
modest degree of rnoling is readily provided if the vacuum case (coupled
radiatively to the rot.or) is t>ooled by wrapping ii with chilled wa.ter pipes. A
more important. benefit of even modest cooling may be the large reduction of
vapor density on axis (by a factor of 45 between 273"K an<l 323°K, if u = 400
m/s) . This rela..xes demands on t.he molPcular pump and decreases sensitivity
to 1na.nufact.uring imperfections that limit its performance. It also reduces
windage loss. Cr<•cp, a significant issue with aluminum (but not steel) rotors,
d1:creases rapidly as temperature is lowered. Cooling may be an indicaLor of
problems with rotor tolerance or of the use of aluminum rotors.
4.3 (U) Importance of Rotor Material Strength
(U//FOUO) The great :;ensitivity to u is apparent in equation (4-2),
~ /fNOFORN~ !DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I ~/ - / NOFORN!tfilQJ 85030-00 7 /09
Page 20
and explains why material streni?,"th (or, more accmatcly, strength-to-wdght
ra.tio) is the most. severely stressed design parameter in gas centrifuges. Cen
tnfuge rotors are made out of high strength-to-weigln materials: maraging
:;teeL aluminum a.Hoy, or (with modern technology) carbon fiber reinforced
polymers. We discuss this further in Sections 4.6 and 5.4.
4.4 (U) Maintaining the Counter-Current Flow
(U //FOHO) The counter-current fl.ow is rising in Lhe interior of the rotor
and descending near its wall. This sense of flow is driven by two effects. acring
together:
• (U //FOUO) Most of thr energy dissipation, typically estimated at
about 30 W per centrifuge , occurs at the bottom as bearing friction
and motor inefficiency. Like water in a pot heated on a stove, the
warmed gas at. th<~ bottom tends to rise. At the rotor wall t,h]s rise
is hindered by viscous friction (a rough estimate shows th~t, for a. gas
prel'\Sure a.t the wall of 0.1 bar. the Reynolds number of countercurrent.
flow l:lt 1 m/sec is roughly 30; this would decre11se in inverse proportion
to the density, but even at, Re= 30 viscous drag is significant) , setting
up an upward flow in the rnterior.
• (U//FOUO) The bottom scoop is immersed in the supersonic azimuthal
flow , and is a significant source of drag. This d rives an Ekman circu
lation very simi lar to the clas.-sical Ekman pumping driven by friction
at the bottom of a stirred vessel (wh ich is why a.znnuthal stirring of a
coffee cup rapidly mixes its contents, and why the fluid rotation damps
much faster than the naive vbcous damping time) . Because the ro
tation at the height of the scoop (or the hottom of the coffee cup) is
slowed by drag, the radial pressure gradient there is reduced. Hence
'ffl-/ 'Bf fNOFORN l1t@J DOE (b)(3)
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I .::r8-/ ~/J<qOFOR1'1~ 85030-007 /09
Page 21
there is a downward vertical pressure force at the rotor wall that drives
fluid downward (from higher layers with a larger centrifugal pres.sure
gradient) imo the slower layer, "'ith a return flow nearer the rotor axis.
The rotating baffi.c separates the countercnrrent flow from the drag of
the upper scoop , which wo11ld tend to drive the flmv in the opposite
direction.
4. 5 (U) Cascade Principles
(U / /FOUO) The quadratic depeudence on the mass difference and the
centripetal a.cceleration (itself proportional to the square of the peripheral
speed) results from these factors coming in twice: once in the sedimentation
force and once in the difference in equilibrium mix fractions.
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ---TSi ~/NOFORN~ 85030-007 /09
Page 22
A well engineered and operated centrifuge could thus
produce separatory work at a rate of about 5 kg-SWU per meter of length.
Significantly lower values ar<~ likely unti l extensive operational experiP-nce is
obtained; Kemp (his Fig. 2.-1) suggests tha.t for a moderate degree of opti
mization 2 kg-SWU/(m-yr) may be realistic .
(ll//POUO) Dirac 's original report is not. eac;ily obtained. A succinct
treatment (and tht~ definitive theory of cascades) is givcu by Cohen [3j. Al
though issued by a major publisher rh1s cau also be hard to find (a xerox
copy is in the .JASON library) . Whitley [4] is the most convenient reference.
but cont.ai us some typogni.phical errors.
4.6 (U) Instabilities and Bellows
(U) An excessively narrow and long centrifuge has very little flexural
stiffness. hence low resonant (critical) frequencies that are 1ndependent of
wall thickness. At fixed v, the rot.ation rate w/27i ex: 1/r. 1f the rotor is to
be operated below the first critical speed (lowest flexural resonance), it. must
have a length-to-diameter ratio
Z/d::; L85(E/a) 114. (4-3)
Ti:Jisfollows from Whitley'sEq . 4 (Part II , p. 70): Z/d $ (1 85/v11'2)(E/p) 11·1•
taking v2 = t:J/p. If v2 < cr/p then the limit on Z/d is somewhat relaxed ,
but efficient separation (oUrrna ex: uJ at coustFLnt E/a and oUma:r ex: ·o3 s at
constant E/p) requires v to be as l~rge as possible.
(l.) It may seem smprising that a result for infinitesimal oscillations of
an elastic tnbe should depeud on its peripheral velocity (gyr~copic effects
-!FSf //NOFORN~
!DOE (b)(3) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
!DOE (b)(3) I ~/ .-;/NOPOR11f ltt!1Q] 85030-007 /09
Page 23
are negligible) , or, eliminating 1J , on its yield strength . The depe!ldcncc on
v arises from usmg Rayleigh's (The Theory of Sound 1877) equation for the
ftexural mode frequency, equi'lting it to the rotational frequency u; = t'/r.
(U //:FOUO) In general, two-dimen:;ional materials are described by
three elastic constants. one for tensile stiffness in each of two principal direc
tions and one for shear (the modulus matrix mfly be diagonalized, with only
two diagonal elements. but the direction of its principal a.'\:es requires a third
parameter to describe). Th is is familiar for ·woven fabrics with fibers i11 two
orthogonal directions· they are st.iff along t.he fiber directions (in general,
with two different stiffnesses if the warp and woof have different numbers of
fibers per unit length . or fibers with different properties) but may he very
compliant in oblique directions if the fibers readily slide over each other .
The tenl)ile stre11gth of carbon fiber reinforced materials in directions oblique
to their fibers (even by sma.ll angles) may be much redrn . .:ed by plastic flow
or era.eking of the matrix and subsequent sliding of tbe fibers. Because the
important parameters are hoop strength and longitudinCll stiffness (with the
separatory (·apa.city proportioual to the 7 /4 power of the first and l/..J power
of t.he ;.;econd}, an optimal tayup might be 1/8 longitudinal fibers and 7 /8
a.zimut.hal fibers. rather than a helical layup.
_!ff;/~ /NOFOrtN IZZBID
!DOE (b)(3) I
!DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -'fflf ~/NOFORN fltBQ] 85030-007 /09
Page 24
(U//FOUO) Uniaxia.I tensile strengths up to 45 kbar (''super high t.en
s11e'' fibf~r) are reported , but , pa.rt1rnliuly because when failure occurs it .is
catast.rophlC, it 1s probably appropriate t.o assume a lesser strength of 30
kbar and modulus of l Mbar (''high t.ensilc. low modulus" fiber: definitions
and estimated properties from wikipedia}. Allowing for t.he weaJi:er epoxy
matrix and the need fo1 fibers in two directious suggests t.hat a conservative
a.<;sur.1pt.ion would be ad>= 10 kbar a.nd Ez: = 1 Mba.r. Then. noting a den
sity of l 8 gm/cm3 , the ma.xi mum v = 745 rn/sec and the limit on subcrit.ical
length Z/d < 5.85. The potential advantage of using rarbon fibers to achei,·e
high v is evident., bul the technology is nm. easy, and the complex anisotropic
11Clt-urc (that may depend on manufacturing details) of these composit.es is
unforgiving in such a ~Mengt.h-critical application .
TS'/ /NOFORN fHi§QJ
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
85030-007 /09 Page 25
5 (U) WHAT ARE THE LONG POLES?
5.1 (U) Precision or exotic machining
(U//FOUO) As a lready ment ioned, Zippe-derived centrifuges a rf' in
most respects a rti fac ts of 1950s machining technology, and can be made in
a machine shop of fairly modest proport ions. Indeed, Zippe's original paper
[2] gives a sense of the scale of t he necessary development effort:
(U //FOUO) [In 1958,] before any technical work could be
started, it was necessary to equip the laboratory in the desired
manner. T he best space avai lable for the work was an old student
cafeteri a bu ilding recently abandoned by t he University. I t. offered
t he promise of adequate space but a complete dismant ling of the
old food-handling equipment and a thorough cleaning were in
order before the installation of fixt ures and services sui table for
IDOE (b)(3) I
IDOE(b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
jDOE (b)(3) I -rs-/~ /NOPOru~lftBQJ
scientific endeavor was possible.
(U//FOUO) A smaJl machine shop has been installed in this
laboratory building. This was suitable for about 90% of the
mechanical work required for the project. The remaining 10%,
usually requiring heavier machinery, has been done either in the
Physics Department shop or in the main shop of the Research
Laboratones for the Engineering Sciences.
(U //FOUO) A room for t.he handling of U F 6 was added later
and a mass spectrometer was inst.ailed.
85030-007 /09 Page 26
IDOE (b )(1 ), (b )(3) I
!DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
!DOE (b)(3) I -".rSf~ /NOFORI~~ 85030-007 /09
Page 27
~ (S)
~ (S)
(U//!i'OUO) These arc the cri t. ical tolerances "to make it work" . Vir
tually all other tolerances need only be good enough "to make it fit" . The
for mer require engineering, t.he latter only craftsmanship.
ISh'RD I (U / /POUO~
-'PS-/ - /NOFORN[HBQJ
IDOE (b)(3) I
100E (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
jDOE (b)(3) I -'PSf ~/NOFORf~ltlffil]
5.1.1 (U) Stationary magnet holder clips
..
~ /NOFORNfltBQ]
IDOE (b)(3) I
85030-007 / 09 Page 28
100E (b)(1) I
IDOE (b)(1}, (b)(3)1
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
5.1.2 (U ) Spiral grooves on the ball bearing
-!FSf jDOE (b)(3) I
85030-007 / 09 Page 29
100E (b)(3) I
100E (b)(1) I
!DOE (b)(1 ), (b)(3) I
Approved for release by OONI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I -'PS-f- /NOFORN fltBQJ 85030-007 /09
Pa,ge 30
-
5.2 (U) Frequency Converters and Motors
( U //FOUO ) Centrifuges require variable-frequency electrical controllers
at pow<:'r levels ,...., 100 W each. and probably less during steady operation:
several centrifugci::; can share a freqnency converter. Frequency converters
draw power from the line power and convert the incoming 50-60 Hz AC volt
age into the variable AC voltages in the 0.1 - 1 kHz range that are required
to drive the hysteresis/synchrouous ccntnfuge motor.
/NOFOR!q~
100E (b)(3) I
-
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I -TS'f~/NOFORN~
~/ !NOFOltN~
!DOE (b)(3) I
85030-007 /09 Page 31
IDOE (b)(1) I
IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I /NOFOltl~ lttfili) 85030-007 /09
Page 32
5.2.1 (U) Commodity audio amplifier chips
(U J/FOUO) The drive frequencic:s required for the motor drive are of
order 0.1 to l kHz a.ud the output power required for the mot.or drive is of
order 100 W . Both of these requirements can be met by generic audio power
amp lifiers, such as the TDA 8950 (NXP Semiconductor: Netherlands) . When
configured as a single-channel bri<lge-tied load, this class-D audio amplifier
can drive :JOOW into an 8 ohm load with less than 0.53 harmonic distortion.
Tt is available for $4.50 in quantities of 100.
5.2.2 (U) Commodity microcontrollers
-TS I ~//NOFORN fltBQ] IDOE (b)(3) I
lDOE (b)(3) I
!DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
!DOE (b)(3) I -'fS-1.~/NOFOR:N!ZtBQJ 85030-007 /09
Page 33
-5.2.3 (U) Commodity power supplies
(U //FOUO) The most stre:;sing requirement of the power supply is to
provide lOOW ar ±36 V to the audio amplifier. Sevcrai generic power supplies
c:a.n be used to rucct this requirement. One example is the SWS300-36 AC
DC converter (TDK-Lambda: Ja.pan), whic:h provides as much as 8 7 A at
36 V DC derived from the AC power line. These AC-DC conveners arc
avail<:Lb)e for $121 in quantities of 10.
(U //FOUOJ All other power rt::qu1rements. such~ t.he 5 \I required w
drive the microcontrollcr , are much Jess stressrng. These caJl either be derived
from the 36 V source using a vollage regulator or independently using r.heir
own AC-DC eonvencr.
5.2.4 (U) Other features
-FS-f ~/NOFORN !ZtBQJ IDOE (b)(3) I
IDOE (b)(3) I
IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
!DOE (b)(3) I ~~/NOPORN~ 85030-007 /09
Page 34
-5.3 (U) Rotor Material and Fabrication
(U//POUO) Here we discuss both the fabrication of the rotors, and the
technology of the maraging steel out of which they are fabricated.
5.3.1 (U) Maraging steel
(U //FOUO) Maraging steel is a martcnsitic preciptation-hardencd very
lo·w carbon steel. It is a commodity item for specialized industrial uses.
mostly aerospace. Suppliers arc easy to locate on the Internet. The word
"maraging" is a contraction of "martensite" and "aging", referring to how it
is made:
• (U / / FOUO) The desired alloy of iron , nickel , and a particular combi
nation of cobalt , molybdenum , titanium, and/or i:iluminum (that will
precipitate as intermeta!lics phases) is smelted in a vacuum furnace.
The nickel , a component. of ordinary st.ainless steels. does not form an
intermet.al lic precipitate, but contributes to corrosion resistance.
TSf-/NOFORN~ !DOE (b)(3) I
IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -'TS-/ f NOFOR!q It@] 85030-007 /09
Page 35
• (U//FOUO) Cast as a preform. it is annealed at 800°C to austenitc.
• (U//POUO) The preform is quenched (air cooicd) to martensit.c. This
produces a relatively soft material that can be cold worked (e.g. , flow
formed) as desired.
• (t: //FOUO) When the final hardening is desired, t he part is "aged"
for 4 hours at 500°C. This causes intcrmet.all ic phases to precipitate ,
giving t.he steel its hardness.
-
-4:&/ I / NOFORN ltt:RQJ IDOE (b)(3) I
!DOE (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -r'S/~fl'JOFORN~ 85030-007 /09
Page 36
(U //FOUO) Ma.raging steel can be v: lded, but must be rr-R.ged after
welding Lo recover its hardnes:;. Possibly the annealing and qucnchmg sr.eps
would also need to be repeated before re-aging; a good metallurgist could <le
t.ermine tliis experiruentaJly in the co11rse of developing any part.irnlar welding
process step.
( U 11FOU8-) Maragmg steel tubes must be passivated to protect them
from the corrosive effects of Uf 6 The standard method Is to use steam
There is an open literature on the subject. (see Slide 39)
5.3.2 (U) Flow forming
-TS-/ //NOFOIU~ fltB:Q] IDOE (b)(3) I
100E (b)(1) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~ ~/NOFORNfHBQJ
5.3.3 (U ) Final hydroforming
"'PS/ /NOFORNltt8Q)
IDOE (b)(3) I
85030-007 / 09 Page 37
100E (b)(3) I
100E (b)(1) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I -.!fSf~/NOPORN~ 85030-007/09
Page 38
-5.4 (U) Bellows
(U //POU 0) Slendei tubes have resonant instabilities that must be
pHsscd through on spin-up (see Section 4.6, above, and Slide lG) . There
a.re two possible way~ to deal with resonances: (1) Always operate subcriti
cally. with rotor speed less than the first resonance. or (2) engineer a method
for passing 5afoly through the first one or two resonances .
(l"f/FOUO) For any given material, material propertie~ dictate the
largest. hP.i.gbt-to-diameter ratio that will support subcritical operation at its
ri1.aximum rotational speed. For a nominal marnging steel 350 with E = 2.1
Mbar. Cf = 350,000 psi = 23.8 khar , p = 8.1 gm/cm3 , v = 540 m/s and
Z./d < 5.67. For aluminum Al-0061-T6. for which E = 68 GPa and u is
a6Ciut 2.4 kbar, Z/d < 7.6. For the higher strcng'Th Al-7075-T6 with E = 72
GPa and er = 4.3 kbar, Z/d < 6.65 at its fast.er limiting speed
(U //FOUO) Zippe introduced bcliows (see Slide 17) to reduce the flex
ural stiffness of the rotor tube for low modes, an<l thus to move their fre
quencies to much lower values. At these lower frequences there b not as
much rotritional energy in the rotor when it passes through r€'sonance, and
the pas.sage can he rna<lc quickly and safely.
-'I's-/~ / !'tOFORH flffi:Q] !DOE (b)(3) I
IDOE (b)(1) I
IDOE (b)(1) I
-
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~~/NOFORN flt[filJ
~(S/~ -5.5 (U) Centrifuge-Specific Know-H ow
5.5.1 (U) Upper Suspension M agnets
'5030-ll07 /09 Page 39
100E (b)(1) I
100E (b)(3) I
100E (b)(3) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I -p&f ~/NOFOR:N ~
5 .5.2 (U) Rotor testing and balancing
85030-007 /09 Paf?;e 40
jDOE (b)(3) I
jDOE (b)(3) I
~.
-=-==:::--:::=== ;.;
-TS-/ ~/NOFORNftlB:Q]
jDOE (b)(3) I
(.S-4)
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I L"S1 ~ / NOFO:ttrq lftB[I 85030-007 /09
Page--!]
-
-( U //FO UOj Accurate balancing is required to ma.iutain clearance, that
might he nominal ly 1 mm, hetwcen the rotor and r.lte molecular pump. The
rotor !:>pins about. it.s center of rnass (the center of mas::. of its cross-sect ion if it
is accurately cylindrical). It the wall thicknes::. were to vary as one sinusoidal
wave around the circ 11rr.ference (a lowest-mode man11facturing imperfection)
with amplitude 6. then its center of mass would depart from its geometrical
axis hy an amount b. = 0.5H.6/w, where R is its radius and 11· its wall
tbickness. lf we demand b. < 0 5 mm and take R = 72.5 mm and w = 0.35
mm. then we require J < 5 µ.
-:PS( ~/NOFORN f@Q] jDOE (b)(3) I
100E (b)(3) I
!DOE (b)(1 ), (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -"f1Sf /NOFORN ~ 85030-007 /09
Page 42
5.5.3 (U) Bearing and damping testing
(l//FOUO) The nat.ural "t1pµing" (in,·erted µendulum) frequency of
the rotor about its lower needle/ball support might be measured rn the stati c:
t.est stand I.JV an oscilliscopc fed by a t.;apacitance gauge. This corresponds
to a vE>ry low frequency resonance that. must be traversed at. the beginning £ .
OL spm-up.
(U//FOUO) The resroring horizontal force is determiner! by the leng1h
of the "dips''. themselves acting as port.ions of a compound pendulum. The
damping of this oscillation is provided by viscou:s dissipation in the oil-filled
/NOFORI>q fftBQ)
!DOE (b)(1 ), (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -'PS-/~ /NOFOffi't flffiQ] 85030-007 /09
Page 43
support . The design sbows screws that can be used to tune the damping by
adjusting t.be thickness of t.he viscous oil layer,
-5.5.4 (U) Secondary flow and test ing
(U//FOUO) Cla.c;sical (subsonic, uniform-density) Aow in Lhe interior of
rotating cylinders supports a. set of flow-instability modes. Arrn.lysis of ttow
in a centrifuge is further complicated by the large radial density gradient ..
The high rnolecular weight, (352) and low adiabaLic exponent (r ::::= 1.07 [·l})
of UFG and its consequent. low speed of sound (at 300°.K 'LIJFc = 87m/s vs.
rta.ir = 347 m/s; cf. 400 m/s circumferential speeds) results in supersonic flow
wit.Ii rei:;µect t,o any stationary :,trucLures. such as t,he product and waste
~oo~s , nedr the rotor. Thus, the Aow pas1 Lhc scoops produces shocks. a.:;
does a pitot t.ubc in supersonic A.ow. that generate shear layers from shock
Cll:\l'lture <:1nd that disLurb ao<l can sepmat.c boundary layers on impiugemcnl
-'P&f ~/NOFORNf@Q] jDOE (b)(3) I
!DOE (b)(1 ), (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -"PSf ~ / I~Ol!'Oftl~ fHBQJ 85030-007 /OIJ
Page -14
with walls Such effects increase mixing that must be minimized . The upper
baffic~ shields the flow from the top scoop
(U //FOUO) In the la.borator.v frame , where nou-rotating cornponent.s
live. secondary-flow ve)ocities are much smaller than the azimuthal velocit.\'.
To a stationary observer, t.he flow in the centrifuge interior resembles a tightly
wound hclb• if no additional instabilities develop. In thn rotating frame ..
t.b e flow is as depicted in Slide 15 of the briefing (Appendix A), except for
complications arising from Coriolis forces on the radial flow at the ends. The
top l.Ja.fflc mitigates a nd isola.tcs effects from tertiary flows (Ekman spirnl)
that are likely to re:-;ult from such forces.
(U tfFOUOr Such flow issues mgue flgainst diagnosing and tuning the
secondary How wit.h intrusive means, such as probes, as these will likely
c~usc significant disturh;:in<:es t.hemsf'lves. The secondary flow could be dia.g
nose <l Ly using surrogal.cs fort.he conos1ve UFt;. While no common surrogate
rna.tches UF6 exactly (352 mol wt , \'apor pressure 1 atm at 56°C), brackct.
ing al different temperatures or rot.ation speeds eould be obt.ained by using
tetrnet.hyl lead (:323 moL wt.. vapor pressure 15 torr at. 84°C) or t.ctramet.h)·l
lead (267 mol. wt., vapor pressure 22.5 torr at 20°C).
{U//FOUO ) Zippe used Freon compounds for his preliminary work a t
the Univcrsit.,v of Virginia but cv0ntt1<'l..lly employed :i mass spectrometer (MS)
to al)sav the sepMative performance of his test centrifuges. This is a re81-
time process that allows optimization of the secondary flow (for example, by
R.dding heat at the bottom m addition to that provided b~- the rnowr drive) .
( 1J //FO UO) In testing a prototype centrifuge it is necessary to know
this "equilibrium time'' --the time f,,, for the centrifuge enriclrnient to :;ettle
ro a new value after a cha.nge in parameters This is the mean residence time
for uFn within the centrifuge and is given by the mass !Vic of UF11 contained
~~//NOFOFL~~ IDOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOJA Case DF-2011-00033.
IDOE (b)(3) I --'PSj ~//NOFORMf@QJ 85030-007 /09
Page 45
divicied by the feed rat~ to the centrifuge, Fr . We will first calculate Fr a11d
then estimate Me for a typical ('(mtrifuge.
('U //FOUO ) Assume a feed rate Fr grams/second to the cent.rifuge that
produces F,./2 of single-stage enriched product of enha.nced in U-235 by a
factor (1 + e) and Fr/2 of smgle-stage product stripped by a factor (1 - e).
The separative work ~U per unit of feed is e2 /2 ([-1] Eq A7) . For a typical
countcrcurrent centrifuge wit.h enr ichment factor 1.05 and stripping factor
U.95 . 6.U is t.h\ls 1.25 x 10- 3 kg-SWU per kg of feed . For a nominal cent.rifuge
with a separative power of 3 kg-SWU f.vr. the feed is thus 2400 kg/yr and F.
= 76 mg/s.
(U //FOUO) For a centrifuge of length Z = 1 m and radius R == 7
c m. the mass of contaiu~d UF 6 i::; the product of the rotor surface area, the
dens ity of UF6 at the rotor (at a µressure le&; t he.n ih; vapor pressure) and its
C<:!ntri fugal ("gravi tation ai .. ) SC"ale height. Scaling to an operating pressure
of 0 .1 atmosphere at the rot.or wall, we find a density at the wall of 1.57
mg/cm3 and a srnle height of about 3.5 mm. Thus /\Jc = 2420 mg.
(U//FOUO] The settling t ime le = ;\fc/ Fr = 32 sec-. This is an upper
limit to le because the separative work output of a centrifuge is independent
of gc:IS denwity; if t.he ma.chine ii; operated at lower densit.y. Fr would rema in
76 mg/s. but 1Wc aIJd le would be correspondingly smaller
(l'//F OUO) In a. test ma.chine. t.hc feed rate would be determined by a
valve set.ting, and t.he operating pressure by valves ill the product and wa.ste
lines to en~ure that it. remain below the vapor pressure at. tbe operat.ing
temperature. The enrichment. would be monitored by mass spectrometer,
perhaps as frequently as every minute, as the heating 1s adjusted to opt1rmze
the secondary circulation. After optimizat.ion. in production centrifuges the
valvet-; might be replaced by constrictions in the feed, product and tni1ing
lines. The heating by the waste heat of the motor cou~d be set by a coating
-PS-/ ~/NOPORN fttBQJ !DOE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-201 1-00033.
witb appropriate emissivity on t.be lower end cap of thf' rotor.
5.6 (U) Manufacturing Scale-Up
--t-8-/ / /NOFORN ftt8Q]
IDOE (b)(3) I
85030-007 /09 Pap,;e 16
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -'PSf ~ / NOFORNfltBll
--
5.7 (U) Human Factors
-'I'S/ ~/i~OFORN!tti§QJ
jDOE (b)(3) I
85030-007 /09 Page 47
lDOE (b)(1) I
IDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
---
5.8 (U) The North Korean Economy
-'!'Sf~ /NOFOR.N fltBQ]
jDOE (b)(3) I
85030-007 /09 Page 48
100E (b)(1) I
100E (b)(3) I
-
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I ..!fSf ~/t<iOFORN lftBQ]
~~/NOFORN~ !DOE (b)(3) I
8.5030-007 /09 PagE> 49
100E (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ""TS-/-/NOFORN~
-".P&f~/NOFORN~ IDOE (b)(3) I
85030-007 /09 Page 50
jooE (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~~/NOFORN fHITQJ
6 (U) ESTIMATED TIMELINES
6.1 (U) How Long Does it Take?
-
4. ~
-'PSf ~ /NOFORN ftl@J IDOE (b)(3) I
85030-007 / 09 Page 51
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -'f'Sf ~ jr-q"OFORNftlBQ]
-
(U/7FOUG-)
-'f'S-/ ~/NOFORtq~
!DOE (b)(3) I
85030-007 /09 Page 52
IDOE(b)(1) I
100E (b)(3) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I 4&f ~ /NOPORNftffiQJ
6.2 (U) Where Are T hey Now?
•
J'.'Sj ~/NOFORP'~ . IDOE (b)(3) I
85030-007 /09 Page 53
100E (b)(3) I
100E (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~~ /NOFORNlftBQJ
7 (U) INDICATORS
7.1 (U) Making Centrifuges
~ ~/NOFORNlttBQ) IDOE (b)(3) I
85030-007 / 09 Page 54
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~ /NOFORI'if~
· ~ • ts? -· ~
~~/NOFORN~ IDOE (b)(3) I
85030-007 /09 Page 55
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -PS/~/NOFORN~
7.2 (U) Operating the Centrifuges
-"fSf ~ /NOFORNlftBQ]
!DOE (b)(3) I
85030-007 /09 Page 56
100E (b)(1) I
IDOE (b)(1) I
-
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
~ ~/NOFOit!vlftBQJ jDOE (b)(3) I
85030-007 /09 Page 57
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I -TS-/~ / NOFOftN·ltfilQJ 85030-007 /09
Page 58
(U) Appendix A: Briefing Slides and Figures
""TSf ~//NOFORN ftl@] !DOE (b)(3) I
JASON 2009
•
Approved forrelease jDOE (b )(3) js. FOIA Case DF-2011-00033. -t5- /tHH Of?N ~
NK Capabilities (U)
JSR-09-510 Summer 2009
~-1 . mr.11.fl ~ IDOE (b)(3) I
North Korea Capabilities
Approved forrelease ~ DOE (b)(3) 16, FOIACase DF-2011-00033
North Korea Centrifuge Capabilities (S)
JASON 2009
JSR-09-510 Summer 2009
Participants:
--·---
~,-N, if e;1?'1~ fltBQ]
!DOE (b)(3) I
!DOE (b)(3), (b)(6) I
North Korea Capabilities 2
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
We thank the government briefers for outstanding, well-organized presentations
JASON 2009
);:- Briefers to JASON
IIIII-
!DOE (b)(3), (b)(6) I
North Korea Capabilities 3
Approved forreleas1DOE (b )(3) r16, FOIA Case DF-2011-00033.
-"ftT,- Jl\I Of OF' M fltffilJ
Summary of what the US knows
• IDOE(b)(1) I
• • • • •
JASON 2009 North Korea Capabilities
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
~f:CRf.. T/tlOt OR~I~
JASON Task Statement (paraphrase)
• JDOE(b)(1) I
•
JASON 2009 North Korea Capabilities 5
I
• I I I •
JASON 2009
Approved forreleas, DOE (b )(3) 116, FOIA Case DF-2011--00033.
~/NOfOl'<N~
T".)1- lfWfOf'N ftt@] IDOE (b)(3) I
North Korea Capabilities
IDOE(b)(1) I
II
II
6
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
FOUU
Manufacturing Best Practice in Industrialized Countries
);:> The fundamental principle is interface control and change control.
~ All parts are fully defined in design by their geometry, materials specifications, and tolerances - 30 CAD always - software verification of fit (e.g. , SolidWorks) - parts are always interchangeable
);:> Design and manufacturing are different engineering disciplines - train engineers to "design for manufacturability" - manufacturer gets to choose the processes while meeting specs - manufacturer gets to change technologies with production rate - bidding on cost encourages production scalability
~ Identical parts can be obtained from different manufacturers - ISO 9000, etc. - certified processes - touch labor and rework are anathemas
~ Properly designed and manufactured parts do not need piece-part testing - Motorola Six Sigma, etc.
""
JASON 2009 North Korea Capabilities
!DOE (b)(3) I
7
• • • • • •
JASON 2009
Approved forreleas1DOE (b )(3) f 16, FOIA Case DF-2011-00033
+Bi- /NUFOt 1'1~
~11u;r,11a ·~
!DOE (b)(3) I
lDOE(b)(1) I
North Korea Capabilities 8
• • •
•
• JASON 2009
Approved for release bjDOE (b )(3) r· FOIA Case DF-2011 -00033.
l'Sfm'NOFORN ~
.........,./ldf'ft)' ' ll~
jDOE (b)(3) I
IDOE(b)(1) I
North Korea Capabilities 9
Approved for release by ODNI on 6/8/2016, FO\A Case DF-2011-00033.
tOUO
The basic ideas (Zippe, 1950s)
Uppe1 )> high g-force makes vacuum on axis (111<.gnoltL)
seals not needed BeA•••u F'aed Wostn Dlolrlbutor - feed on axis
)> counter-current creates large bottom-to-top Top E1u.1Cnv gradient
Mole,Ul;;i Pump - no delicate scooping within a scale height of
Casuig Top material Scoop
)> magnetic bearing naturally allows Balfto (rotalmg) acceptable wobble without damage
- and takes load off needle bearing
)> "molecular pump" is simply a machined ConterDOst helix in the case (•l~llonary) Bottom
Sroop - no moving parts
/I l:lollom )> motor uses bottom end cap as its magnetic F.11d C"p rotor vacuuin
Pump
)> bellows (not shown) allow operation above 1 or 2 resonances
£ivr. troma9nelic - move them to low speeds MolOJ'
JASON 2009 rouo North Korea Capabilities 10
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
FOUtJ
The physics is elegant
JASON 2009 North Korea Capabilities 11
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
bE:Cl.t; t!tWI Ofifl lttBQ]
IDOE(b)(1) I
JASON 2009 North Korea Capabilities 12
Approved for release by ODNI on 61812016, FOIA Case DF-2011-00033.
I UUU
The cascade principle is used both within a single 100E (b)(3) 1centrif uge and in the cascade of multiple centrifuges
- .. ., __ ~ + _ _ ... .-·-~ ------ --
' ,
Since the cascade is exponential, its length is approximately a logarithm. Each stage is an appropriate number of centrifuges in parallel.
nlimuer o stages - n · --· n .s ,,.,,., HEU'=" 92% L • f . . _ l ( 1 - J tail /prod ) I l ( ) fta.il 1 - J prod "'
ti· -P' % tails Qqo/.,- ln[l/0.4%] + ln[l/(1 - 92%)] ::~ - ~ 40
JASON 2009
0: 1 1 n,~llmcnt
per r:ontrifu•JO
FtltJG
0.2
North Korea Capabilities 13
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 14
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
f \,.!ULI
JASON 2009 North Korea Capabil ities 15
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
l vUtJ
Bellows (1)
JASON 2009 North Korea Capabilities 16
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
;:,(:'.Ht f 1''40! Uf' I ' ltiB:QJ
JASON 2009 North Korea Capabilities 17
JASON 2009
Approved forrelea1DOE (b )(3) ~016 , FOIA Case DF-2011-00033.
-+.;-l~OF0 1 'l'i ~
"'t'""~f<ft•I Ql.f~ ltfilQJ
IDOE (b)(3) I
IDOE (b)(1), (b)(3) I
North Korea Capabilities 18
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 19
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
SLtHLTtrllOI ogr; ~
IDOE (b)(1)1
JASON 2009 SECht::1tlWf U!'t1~ North Korea Capabilities 20
Approved for release 1DOE (b )(3) 16, FOIA Case DF-2011-00033.
--ftr,
Required processes add up to machine-shop scale 100E (b)(1).(b)(3) 1
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
1 tJUO
Zippe's original paper gives a sense of the scale
[ln 1958, ] before any technical work could be started, it was necessary to equip the laboratory in the desired manner_ The best space available for the work was an old student cafeteria building recently abandoned by the University . It offered the promise of adequate space but a complete dismantling of the old food-handling equipment and a thorough cleaning were in order before the installation of fixtures and services suitable for scientific endeavor was possible.
A small machine shop has been installed in this laboratory building. This was suitable for about 90% of the mechanical work required for the project. The remaining 10%, usually requiring heavier machinery , has been done either in the Physics Department shop or in the main shop of the Research Laboratories for the Engineering Sciences.
A room for the handling of UF6 was added later and a mass spectrometer was installed.
JASON 2009 North Korea Capabilities 22
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 23
Approved for releastooilt(b)('3i12l 16, FOIA Case DF-2011-00033.
~-/l'jtJl"0ji'ft1 ~
JASON 2009 ~1-.. , . , , , · 11 , 11~ !DOE (b)(3) I North Korea Capabilities 24
Approved forrelease jDOE {b )(3) 16, FOIA Case DF-2011-00033.
4---NOJ <)i'"~
JASON 2009 North Korea Capabilities 25
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
!DOE (b)(3) I
JASON 2009 28
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011--00033.
JASON 2009 North Korea Capabilities 29
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
rouo
Software workbench for the microcontroller is a commodity item
JASON 2009 North Korea Capabilities 30
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
:':>ECF<ETlNOFOf'ti ~
Rotor material and fabrication?
JASON 2009 Sf::P.ET!fJOFQP · ~ North Korea Capabilities 31
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
SECfU::l/110! ORld ~
JASON 2009 North Korea Capabilities 32
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
FOUO
..
(play movie)
JASON 2009 iOUO North Korea Capabilities 33
JASON 2009
Approved for release by ODNI on 6/8/2016, FOIA Case DF-20.11-00033.
FO!Jt;:.1
Maraging steel is a commodity item for specialized industrial uses
North Korea Capabilities 34
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
Maraging steel basics
JASON 2009 n .. uu North Korea Capabilities 35
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
36
Approved for release by ODNI on 61812016, FOIA Case DF-2011-00033.
37
JASON 2009
Approved for releas, DOE (b )(3) ]1 6, FOIA Case DF-2011-00033.
~t'NOiOHf~~
~-i·hJtl)! ··~ lDOE (b)(3) I
North Korea Capabilities 38
Approved for release by OONI on 6/8/2016, FOIA Case OF-2011-00033.
,;>f.CHL i IUOI ORN~
I !DOE (b)(3) I
I
I
I
!DOE (b)(6) I JASON 2009 North Korea Capabilities 39
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE(b)(1) I
JASON 2009 North Korea Capabilities 40
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 4 1
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 42
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011--00033.
JASON 2009 North Korea Capabilities 43
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
44
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE(b)(1) I
JASON 2009 North Korea Capabilities 45
Approved forrelease bl DOE (b )(3) r· FOIA Case DF-2011-00033.
~·-NvttJ•'fl~
JASON 2009 North Korea Capabilities 46
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
SECRFT/NOFORN~
Timelines and indicators
JASON 2009 :St:Cf~L fJ'NOFQf<.t t~ North Korea Capabilities 47
JASON 2009 ~ ... ltt). t>fN~ !DOE (b)(3) I
North Korea Capabilities 48
JASON 2009
Approved for releaselDOE (b )(3) 11 6, FOIA Case DF-2011-00033.
~iN!.)I Ui'N~
~1Nti10f~'~
!DOE (b)(3) I North Korea Capabilities 49
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 50
JASON 2009 North Korea Capabilities 51
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
JASON 2009 North Korea Capabilities 52
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I
Ts./~/NOFORN INRD I
(U) Appendix B: CHOP
-
-'f'S'f ~/NOFORN" IHRD I IDOE (b)(3) I
85030-007 /09 Page 110
IDOE(b)(1) I
-
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I
-r.rsf ~/NOFORNIHRD I
"':PB-/ ~/NOFOru-tlt'tRD I IDOE (b)(3) I
85030#007 /09 Page 111
IDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I ~/ ~ /NOFORM lh'RD I
-r&/~/NOFORN lnRD I !DOE (b)(3) I
85030·007 /09 Page 112
IDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I .!ffi/ //:NOFORl't IHRD I 85030-007 /09
Page 113
(U) Appendix C: ,..rowards Better Estin1ates of Program Size
-
4:1Si~//NOFORN IHRD I
jDOE (b)(3) I
100E (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I ~~/I~OFOruq lffRD I
(U //FOUO) See [1] for discussion related to this section
~~/NOFORNlh'RD I !DOE (b)(3) I
85030-007 /09 Page 114
IDOE (b)(1) I
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
IDOE (b)(3) I "'PSf ~/ /NOFORr.qlh'f~D I 85030-007/09
Page 115
(U) Appendix D: Size of an Enrichment Cascade Facility
IDOE (b)(1) I
ISHNFI
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011-00033.
100E (b)(3) I
-&fS-/ ~ /NOFORNIHRD I
~/~/NOFORN l/tRD I IDOE (b)(3) I
85030-007 /09 Page 116
~ID-OE_(_b )-( 1 ~) I
Approved for release b1DOE (b )(3) J' FOIA Case DF-2011-00033.
"""fSi'
Appendix E: Motor Windings
-
·-· •••• -fSt'- /NQVOR"N lllRD I
!DOE (b)(3) I
135030-007/09 Page 116
IDOE (b)(3) I
Approved for release !DOE (b )(3) ,6. FOIA Case DF-2011-00033.
-<ffif- /fNOf'OftN llfF~D I
-'PS/
IDOE (b)(3) I
85030-007/09 Page 117
IDOE (b)(3) I
Approved for release I DQE (b )(3) r· FOIA Case DF-2011 -00033.
--ffl;'~A>10FORN IHRD I
ffi- /NOFORlq IHRD I jDOE (b)(3) I
85030-007/09 Page 118
Approved for releaselDOE (b )(3) 116, FOIA Case DF-2011-00033.
~//NOfORN IHRD I 85030-007/09 Page 119 IDOE (b)(3) I
ICHRDI (U)-·
•••••
-TS-/
!DOE (b)(3) I
-
Approved for release joO'E (b}'(3)'f' FOIA Case DF-2011-00033.
4"St_.11c;10-rom IHRD I 8so30-001109
~moroRN IHRD I IDOE (b)(3) I
!DOE (b)(3) I Page 120
,_
jCllRDI
Approved for release by ODNI on 6/8/2016, FOIA Case DF-2011 -00033.
IDOE (b)(3) I ~ - /NOFORNl//RD I
References
I
I I
'11S/ ~ ;roqo:rrORT~ IHRD I jDOE (b)(3) I
85030-007 /09 Page 121
IDOE (b)(3) I