Antineutrino reactor safeguards - a case study
AAntineutrino reactor safeguards – a case study
Eric Christensen, Patrick Huber † , Patrick Jaffke Center for Neutrino Physics, Virginia Tech, Blacksburg, VA 24061, USA
Abstract
Antineutrinos have been proposed as a means to safeguard nuclear reactors for morethan 30 years and there has been impressive experimental progress in antineutrinodetection that makes this method increasingly practical for use by the InternationalAtomic Energy Agency. In this paper we conduct, for the first time, a case study ofthe application of antineutrino safeguards to a real-world scenario – the North Koreannuclear crisis in 1994. We derive detection limits to a partial or full core discharge in1989 based on actual IAEA safeguards access and find that two independent methodswould have yielded positive evidence for a second core with very high confidence. Togeneralize our results, we provide detailed estimates for the sensitivity to the plutoniumcontent of various types of reactors, including most types of plutonium productionreactors, based on detailed reactor simulations. A key finding of this study is that awide class of reactors with a thermal power of 0.1-1 GW th can be safeguarded achievingIAEA goals for quantitative sensitivity and timeliness with antineutrino detectors rightoutside the reactor building. This type of safeguards does not rely on the continuityof knowledge and provides information about core inventory and power status in real-time. † Email: [email protected] a r X i v : . [ phy s i c s . i n s - d e t ] F e b ontents e reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.2 IRT reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335.3 5 MW e reactor power measurement at IRT . . . . . . . . . . . . . . . . . . . 355.4 Waste detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375.5 Continuous neutrino observations . . . . . . . . . . . . . . . . . . . . . . . . 385.6 Impact of backgrounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 e reactor 59B CANDU and LEU reactors 62 The IRT reactor 63 Introduction
The first use of nuclear weapons in 1945, at the end of World War II, had a profound andpermanent impact on foreign relations and international security. While initially there wassome hope that the secrets of the manufacture of nuclear weapons would remain exclusivelyin the hands of the United States, the Soviet Union tested its first nuclear device in 1949.Several times during the Cold War, the world stood at the brink of nuclear armageddon.It was only due to the strong commitment of political leaders on both sides and the highdegree of professionalism in the armed forces that the disaster of a nuclear war could beaverted . During the Cold War, nuclear security was essentially a bipolar issue between theUnited States and the Soviet Union; other players like Great Britain (1952), France (1960),and China (1964) appeared at the fringes but did not play a major role for most parts. Tosome degree independently of Cold War politics, Israel and South Africa launched a nuclearweapons program in the early 60s and 70s, respectively, which in both cases was triggered byunique national security needs – a small minority population surrounded by hostile neigh-bors, which in turn resulted in a rather unusual alliance. South Africa, quite remarkably,relinquished its nuclear weapons and the associated infra-structure towards the end of theapartheid regime in 1991. India’s nuclear program was launched quite early and presumablywas a direct response to its deteriorating relations with Pakistan. Also confrontations withChina over territories in the Himalayas in combination with China obtaining a permanentseat on the UN Security Council contributed significantly to the decision to go nuclear. Ob-viously, India’s possession of nuclear weapons since 1974, then created a perceived need fornuclear armament in Pakistan, which first tested a nuclear device in 1998. The last countryto join the circle of nuclear armed nations was the Democratic People’s Republic of Korea
1. Thomas Reed,
At the Abyss: An Insider’s History of the Cold War (Presidio Press, 2004). For adifferent perspective, why we the Cold War remained cold, see for instance Scott D. Sagan,
The Limits ofSafety (Princeton University Press, 1993).2. Sasha Polakow-Suransky,
The Unspoken Alliance: Israel’s Secret Relationship with Apartheid SouthAfrica (Pantheon, 2010).3. James Doyle, ed., “Nuclear Safeguards, Security and Nonproliferation: Achieving Security with Tech-nology and Policy,” in (Butterworth-Heinemann, 2008), chap. 16. The NPT is, with exception of the UN charter, the most widely accepted internationaltreaty to date. Currently, 190 states are party to the NPT. The legal mechanisms for IAEAsafeguards, as set out in article III of the NPT, are bi-lateral agreements between individualmember states and the IAEA. These so-called comprehensive safeguards agreements havebeen put into force by all but 12 of the non-nuclear-weapons states. The Additional Protocolwas introduced in response to the failure of the regular safeguards scheme to provide timelyindication of Saddam Hussein’s nuclear weapons program before the first Gulf War in 1990.The Additional Protocol in particular provides IAEA inspectors with the right to collectenvironmental samples at locations outside of declared facilities and to obtain access tosites which have not been declared as nuclear facilities but are suspected to be. Theseprovisions close an important gap in the regular safeguards scheme, which relies on a state’sdeclaration of nuclear facilities and materials. 139 states have signed the Additional Protocol,and 117 states have put it into force. The regular safeguards scheme could only confirmthe correctness of a state’s declaration of nuclear activities; with the Additional Protocol,the completeness of the declaration can also be addressed . It can be argued that thecorrectness aspect of safeguards is working quite well. No case of diversion of fissile materialhas been documented at safeguarded facilities, which presumably is due to the fact that apotential proliferator deems the risk of discovery to be unacceptably high. However, thecompleteness aspect remains troubling, especially for those states which have not put the
4. “NPT,” (accessed November 19,2013).5. “NPT status,” (accessed December 18, 2013).6. “IAEA webpage,” (accessed Septem-ber 13, 2012).7. In principle, special inspections could also provide the means to verify the completness, independent ofwhether the Additional Protocal is in force.8. Sergey Zykov, “IAEA Instrumentation for the Future,” in
Proceedings of the 53rd Annual Meeting ofthe Institute for Nuclear Materials Management (INMM) (2012). in 1956 using neutrinos from the SavannahRiver reactor. Neutrinos are nearly massless, electrically neutral, spin 1/2 particles and playa central role in the electroweak Standard Model of particle physics. Neutrinos participateonly in weak interactions and therefore possess unusual penetrating power – no practicalmeans to attenuate or to shield neutrinos are known. Neutrinos are copiously producedin the beta-decays of fission fragments and this makes nuclear reactors the most powerfulartificial neutrino source. The basic concept to monitor nuclear reactors using neutrinos wasproposed by Borovoi and Mikaelyan in 1978. There have been a number of quantitativestudies of the level of accuracy at which the plutonium content in a reactor can be deter-mined using neutrinos, and different authors seem to come to different conclusions. Closerinspection of those results reveal that very different assumptions about detector capabilitiesare made and also the level of statistical analysis, particularly in terms of rates versus spec-tral information, is very different. These differences likely account for the variety of opinionson the feasibility and quantitative accuracy of neutrino safeguards. In particular, the as-sumptions about detector capabilities seem to be strongly influenced by earlier safeguardsdetector deployments and do not reflect modern state-of-the-art neutrino detectors. We will
9. C. L. Cowan et al., “Detection of the free neutrino: A Confirmation,”
Science
124 (1956): 103–104,doi: .10. To be precise, a reactor is a source of electron antineutrinos. In the interest of brevity and given thefact that here we are dealing only with electron antineutrinos, we will use the term neutrino, instead.11. A. A. Borovoi and L. A. Mikaelyan, “Possibilities of the practical use of neutrinos,”
Soviet AtomicEnergy
44 (1978): 589.12. Adam Bernstein et al., “Nuclear reactor safeguards and monitoring with anti-neutrino detectors” (2001);Michael Martin Nieto et al., “Detection of anti-neutrinos for nonproliferation” (2003); Patrick Huber andThomas Schwetz, “Precision spectroscopy with reactor anti-neutrinos,”
Phys. Rev.
D70 (2004): 053011,doi: ; A.C. Misner, “Simulated Antineutrino Signatures of Nuclear Reactorsfor Nonproliferation Applications” (PhD diss., Oregon State University, 2008); A. Bernstein et al., “NuclearSecurity Applications of Antineutrino Detectors: Current Capabilities and Future Prospects,”
Sci.GlobalSecur.
18 (2010): 127–192; Vera Bulaevskaya and Adam Bernstein, “Detection of Anomalous Reactor Ac-tivity Using Antineutrino Count Rate Evolution Over the Course of a Reactor Cycle,”
J.Appl.Phys. (2010);A.C. Hayes et al., “Theory of Antineutrino Monitoring of Burning MOX Plutonium Fuels,”
Phys.Rev.
C85(2012): 024617, doi: ; Patrick Huber, “Spectral antineutrino signatures andplutonium content of reactors,” in
Proceedings of the 53rd Annual Meeting of the Institute for Nuclear Ma-terials Management (INMM) (2012). .Here, we present the first case study of a real safeguards scenario – the first nuclear crisisin the DPRK in 1994. Earlier studies of neutrinos for safeguards are, to a large degree, basedon pre-conceived notions of how safeguards of a particular reactor type work, and thus do notallow critical examination of the strength and weaknesses of neutrino safeguards as comparedto more conventional means. In many cases, this comparison seems to disfavor neutrinos, notbecause neutrinos do not offer any new capabilities, but because the conventional techniquesare specifically designed to work well in those standard scenarios. On the other hand,inventing scenarios in which the standard methods fail brings about the criticism that thesescenarios are artificial, unrealistic, and contrived for the sole purpose of demonstrating theusefulness of neutrinos. In the rare case that any of these scenarios would reflect an actualconcern of the professional safeguards community, it is far from obvious that anyone in thatcommunity would want to admit it. Therefore, what is needed is a real-world case in whichconventional methods did not yield the desired outcome for the IAEA and for which sufficientinformation is publicly available to perform a detailed technical analysis. The first NorthKorean nuclear crisis fits this bill on all accounts and we, therefore, have chosen it as oursandbox to explore the abilities and limitations of neutrino safeguards. Despite the briefdiscussion of the impact neutrino safeguards might have had on the unfolding of history insection 6.2, the main thrust of this study is not an attempt at counter-factual history but todemonstrate that under real-world constraints and boundary conditions, neutrino safeguardscan provide a decisive advantage over conventional techniques, in particular, with a view ofthe next nuclear crisis in some other part of the world.This paper is organized as follows: the technical aspects of neutrino safeguards and the
13. for a recent review on this topic, see Glenn R. Jocher et al., “Theoretical antineutrino detection,direction and ranging at long distances,”
Phys.Rept.
527 (2013): 131–204, doi:
14. Joel S. Wit, Daniel Poneman, and Robert L. Gallucci,
Going Critical: The First North Korean NuclearCrisis (Brookings Institution Press, 2007).15. Zykov, “IAEA Instrumentation for the Future.”
The fact that nuclear reactors are powerful neutrino sources was realized soon after nuclearreactors became practical. Neutrinos are not directly produced in nuclear fission but resultfrom the subsequent beta-decays of the neutron-rich fission fragments. On average there areabout 6 neutrinos per fission emitted and thus, for one gigawatt of thermal power a flux ofabout 10 s − neutrinos is produced. The total number of emitted neutrinos is proportionalto the total number of fissions in the reactor. Moreover, the distribution of fission fragments,and hence their beta-decays, are different for different fissile isotopes. Thus, careful neutrinospectroscopy should provide information not only on the total number of fissions but alsoabout the fission fractions of the various fissile isotopes contained in the core. The basicconcepts of both power monitoring and observing the plutonium content of a reactor wereexperimentally demonstrated in pioneering work performed by a group from the KurchatovInstitute lead by Mikaelyan. They deployed a neutrino detector of about 1 m volume at the
16. Borovoi and Mikaelyan, “Possibilities of the practical use of neutrinos.” and the effect due to a changing plutonium contentwas demonstrated; more recently the quantitative accuracy has been studied as well. Thisallows one to determine the plutonium content and power level of the reactor core in situ ata standoff distance of 10’s of meters. The practical feasibility of reactor monitoring usingneutrinos has also been demonstrated using a small, tonne-size detector at the San Onofrepower station, called SONGS. Beginning with the discovery of the neutrino, inverse beta-decay (IBD) has been the workhorseof reactor neutrino experiments ¯ ν e + p → n + e + (1)In IBD, an electron antineutrino interacts with a proton to produce a neutron and a positron.Due to the mass difference of a neutron and a proton as well as the mass of the positron, thisprocess has an approximate energy threshold of ( m n − m p + m e ) c = 1 . E e and thus the visible energy in detector, E vis = E e + 2 ×
511 keV. There is a one-to-onecorrespondence between neutrino energy and the positron energy E ν = E e + 1 . E vis directly translates into
17. V. A. Korovkin et al., “Measuring Nuclear Plant Power Output by Neutrino detection,”
Soviet AtomicEnergy (1988): 712–718.18. Yu. V. Klimov et al., “Measurement of variations of the cross section of the reaction ¯ ν e + p → e + + n in the ¯ ν e flux from a reactor,” Sov. J. Nucl. Phys.
51, no. 2 (1990): 225–258.19. Huber and Schwetz, “Precision spectroscopy with reactor anti-neutrinos.”20. Yu. A. Klimov et al., “Neutrino method remote measurement of reactor power and power output,”
Atomic Energy
76, no. 2 (1994): 123–127; Huber, “Spectral antineutrino signatures and plutonium contentof reactors.”21. A. Bernstein et al., “Monitoring the Thermal Power of Nuclear Reactors with a Prototype Cubic MeterAntineutrino Detector,”
J.Appl.Phys.
103 (2008): 074905, doi: . E ν .The reaction in equation 1 also results in a neutron, which in itself is invisible to thedetector, but will slow down in collisions with the detector material and eventually undergoneutron capture. A careful choice of the nucleus on which the neutron captures allowstailoring this signature. Common neutron capture agents are gadolinium, e.g. Daya Bay or lithium, e.g. Bugey. In the case of gadolinium, the signature of neutron capture is theemission of several gamma rays with a total energy of 8 MeV, whereas in the case of lithiumthe signature is the production of an alpha particle and a H nucleus. The slowdown andcapture of the neutron requires a characteristic time, allowing for what is called a delayedcoincidence: there is a primary energy deposition from the positron followed somewhat laterby a neutron capture signal. This delayed coincidence is key to separate neutrino events frombackgrounds. The neutron capture cross sections of both gadolinium and lithium are muchlarger than of any of the other detector materials. Therefore, even small concentrations ata level of a percent or less will result in the majority of neutron captures occurring on thosenuclei.Eventually, all signatures will result in ionization and this ionization is detected by usingorganic scintillator which can be either liquid or solid. The organic nature of the scintillatorprovides the free protons for the interaction in equation 1. Recently, there have been threeexperiments aimed at fundamental physics employing gadolinium-doped liquid scintillatorat a large scale of several 10 tonnes without any safety incidents and excellent long-termstability. Specifically, throughout this paper we consider a 5 t detector based on organicscintillator corresponding to 4 . × target protons. A real detector will not have 100%efficiency and to obtain the same number of events a larger detector will be needed. Many
22. F.P. An et al., “Observation of electron-antineutrino disappearance at Daya Bay,”
Phys.Rev.Lett. .23. Y. Declais et al., “Search for neutrino oscillations at 15-meters, 40-meters, and 95-meters from a nuclearpower reactor at Bugey,”
Nucl.Phys.
B434 (1995): 503–534, doi: .24. Y. Abe et al., “Indication for the disappearance of reactor electron antineutrinos in the Double Choozexperiment,”
Phys.Rev.Lett.
108 (2012): 131801, doi: ; An et al., “Obser-vation of electron-antineutrino disappearance at Daya Bay”; J.K. Ahn et al., “RENO: An Experiment forNeutrino Oscillation Parameter θ Using Reactor Neutrinos at Yonggwang” (2010). and a net load capacity of28.2 t, thus even a 10 t neutrino detector fits easily within such a container together withits support systems. The neutrino spectrum is divided in energy from 1.8 MeV to 8 MeV inbins of 0.2 MeV width, which at 4 MeV approximately corresponds to 10% / √ E resolution,which is similar to the resolution of recent experiments. We checked that a resolution halfas good would yield virtually identical results. For the IBD cross section we use the resultof Vogel and Beacom corrected for a neutron lifetime of 878.5 s . For all measurementsat reactors, the standoff is 20 m, which for both of the considered reactors would allow fordeployment outside the reactor building. Such a detector at this standoff would typicallyregister about 5,000 events per year for a reactor operating at 1 MW th throughout that year. More than 99% of the power in reactors, in a uranium fuel cycle, is produced in the fissionof four isotopes: uranium-235, plutonium-239, uranium-238, and plutonium-241. A reactorwith fresh fuel starts with only fissions in the uranium isotopes and plutonium is producedvia neutron capture on uranium-238 as the burn-up increases. The total neutrino flux froma reactor φ can be written as φ ( E ) = (cid:88) I f I S I ( E ) , (2)
25. Abe et al., “Indication for the disappearance of reactor electron antineutrinos in the Double Choozexperiment”; An et al., “Observation of electron-antineutrino disappearance at Daya Bay”; Ahn et al.,“RENO: An Experiment for Neutrino Oscillation Parameter θ Using Reactor Neutrinos at Yonggwang.”26. P. Vogel and John F. Beacom, “Angular distribution of neutron inverse beta decay,”
Phys.Rev.
D60(1999): 053003, doi: .27. This values is taken from () and is very close to the value of 880 s currently recommended by the ParticleData Group,. () It should be mentioned that there still are measurements deviating significantly from thatvalue by several standard deviations, see e.g.. () f I is the fission rate in isotope I and S I ( E ) is the neutrino yield for the isotope I .The thermal power of the reactor is also given in terms of the fission rates P th = (cid:88) I f I p I , (3)where p I is the thermal energy release in one fission of the isotope I ; we use the values for p I given by Kopeikin. In order to be able to disentangle the contributions of the four isotopes,we need to know the neutrino yields S I . These neutrino yields, in principle, are given by theneutrino spectra ν k ( E ) of each fission fragment k and the cumulative fission yield for eachfragment, Y Ik , S I ( E ) = (cid:88) k Y Ik ν k ( E ) , (4)where k typically runs over about 800 isotopes. In practice, we do not know the neutrinospectrum of a given fission fragment, but have only information regarding the beta spectrumand in many cases this knowledge is inaccurate, incomplete, or entirely missing. Even fora well known beta spectrum, significant complications arise from the conversion of a betaspectrum into a neutrino spectrum since each individual beta-decay branch has to be treatedseparately. As a result, a direct computation of the neutrino yields S I via the summation ofall individual neutrino spectra will be of limited accuracy, but in many cases is the onlyavailable method.A more accurate method is based on the measurement of the integral beta spectrum of allfission fragments and subsequently the neutrino spectrum can be reconstructed from those
28. V. Kopeikin, L. Mikaelyan, and V. Sinev, “Reactor as a source of antineutrinos: Thermal fission energy,”
Phys.Atom.Nucl.
67 (2004): 1892–1899, doi: .29. Th. A. Mueller et al., “Improved predictions of reactor antineutrino spectra,”
Phys. Rev. C
83, no.5 (2011): 054615, doi: ; M. Fallot et al., “New antineutrino energy spectrapredictions from the summation of beta decay branches of the fission products,”
Phys.Rev.Lett.
109 (2012):202504, doi: .30. F. Von Feilitzsch, A.A. Hahn, and K. Schreckenbach, “Experimental beta spectra from Pu-239 andU-235 thermal neutron fission products and their correlated anti-neutrino spectra,”
Phys.Lett.
B118 (1982):162–166, doi: ; K. Schreckenbach et al., “Determination of the anti-neutrinospectrum from U-235 thermal neutron fission products up to 9.5 MeV,”
Phys.Lett.
B160 (1985): 325–330,doi: ; A.A. Hahn et al., “Anti-neutrino spectra from Pu-241 and Pu-239 This method is less dependent on nuclear data about individual fissionfragments but is not entirely free from uncertainties related to effects of nuclear structure. We need to point out that the problem of neutrino yields has recently received significantscrutiny. Until the 2011 work by a group from Saclay, the results by Schreckenbach et al., obtained in the 1980s at the Institut Laue-Langevin in Grenoble were considered the goldstandard. The Saclay group, in preparation of the Double Chooz neutrino experiment, revisited the previous results in an attempt to reduce the uncertainties. Instead, they founda upward shift of the central value of the average yield by about 3% while the error budgetremained largely unchanged. This result, in turn, requires a reinterpretation of a largenumber of previous reactor neutrino experiments, since this changes the expected numberof events. Together with the changes of the value of the neutron lifetime and correctionsfrom so-called non-equilibrium effects, the previous experiments appear to observe a deficitin neutrino count rate of about 6%; this is called the reactor antineutrino anomaly andwas first discussed by Mention et al.. The initial result on the flux evaluation and the 3%upward shift was independently confirmed by one of the authors. A plausible explanationcould come in the form of a new particle, a sterile neutrino, which is not predicted by theStandard Model of particle physics. Given the far-flung consequences of the existence of this thermal neutron fission products,”
Phys.Lett.
B218 (1989): 365–368, doi: ;N. Haag et al., “Experimental Determination of the Antineutrino Spectrum of the Fission Products of
U”(2013).31. Patrick Huber, “On the determination of anti-neutrino spectra from nuclear reactors,”
Phys.Rev.
C84(2011): 024617, doi: .32. Huber, “Anti-neutrino spectra”; A.C. Hayes et al., “Reanalysis of the Reactor Neutrino Anomaly”(2013).33. Mueller et al., “Improved predictions.”34. Von Feilitzsch, Hahn, and Schreckenbach, “Experimental beta spectra from Pu-239 and U-235 thermalneutron fission products and their correlated anti-neutrino spectra ”; Schreckenbach et al., “Determinationof the anti-neutrino spectrum from U-235 thermal neutron fission products up to 9.5 MeV ”; Hahn et al.,“Anti-neutrino spectra from Pu-241 and Pu-239 thermal neutron fission products .”35. Abe et al., “Indication for the disappearance of reactor electron antineutrinos in the Double Choozexperiment.”36. Fred E. Wietfeldt and Geoffrey L. Greene, “
Colloquium : The neutron lifetime,”
Rev. Mod. Phys.
Phys.Rev.
D83 (2011): 073006, doi: .38. Huber, “Anti-neutrino spectra.” .In table I the event rate predictions for various flux models are compared for the four fissileisotopes. The ENSDF flux model is based on thermal neutron fission yields of uranium-235,plutonium-239, and plutonium-241 from the JEFF database, version 3.1.1; the fast neutronfission yield of uranium-238 from the ENDF-349 compilation conducted at Los Alamos Na-tional Laboratory; and on the beta-decay information contained in the Evaluated NuclearStructure Data File (ENSDF) database, version VI. The neutrino spectrum is computedfollowing the prescription of Huber. Our ENSDF model represents a very crude summa-tion calculation and we reproduce the measured total beta spectra to within about 25%.A state-of-the-art summation calculation is given by Fallot et al. , where great care is takento replace the ENSDF entries with high quality experimental data where available and touse a carefully selected mix of databases. This model reproduces the measured total betaspectra to within 10%. Finally, a direct deconvolution of the neutrino spectra from thetotal beta data was performed by Huber for the isotopes uranium-235, plutonium-239, andplutonium-241, which to this date represents the most accurate neutrino yields for thoseisotopes. We note that the absolute values differ significantly between models, but once wenormalize the predictions for total rate and mean energy to that of uranium-235, the predic-
39. for a recent review, see K.N. Abazajian et al., “Light Sterile Neutrinos: A White Paper” (2012)40. “JEFF database,” .41. T. R. England and B.F. Rider,
ENDF-349 Evaluation and Compilation of Fission Product Yields: 1993 ,LA-UR-94-3106, technical report (Los Alamos National Laboratory, 1994).42. “ENSDF database,” .43. Huber, “Anti-neutrino spectra.”44. Von Feilitzsch, Hahn, and Schreckenbach, “Experimental beta spectra from Pu-239 and U-235 thermalneutron fission products and their correlated anti-neutrino spectra ”; Schreckenbach et al., “Determination ofthe anti-neutrino spectrum from U-235 thermal neutron fission products up to 9.5 MeV ”; Hahn et al., “Anti-neutrino spectra from Pu-241 and Pu-239 thermal neutron fission products ”; Haag et al., “ExperimentalDetermination of the Antineutrino Spectrum of the Fission Products of
U.”45. Fallot et al., “New antineutrino energy spectra predictions from the summation of beta decay branchesof the fission products.”46. Von Feilitzsch, Hahn, and Schreckenbach, “Experimental beta spectra from Pu-239 and U-235 thermalneutron fission products and their correlated anti-neutrino spectra ”; Schreckenbach et al., “Determination ofthe anti-neutrino spectrum from U-235 thermal neutron fission products up to 9.5 MeV ”; Hahn et al., “Anti-neutrino spectra from Pu-241 and Pu-239 thermal neutron fission products ”; Haag et al., “ExperimentalDetermination of the Antineutrino Spectrum of the Fission Products of
U.”47. Huber, “Anti-neutrino spectra.” able I: Rates and mean energies (cid:104) E (cid:105) for a 1 MW th reactor in a 1 t detector at a standoff of 10 mmeasuring for 1 year for each individual isotope, assuming that only this isotope is fissioning. Thethree different flux models are explained in the text. Ratios are given relative to uranium-235.ENSDF Fallot Huberrate (cid:104) E (cid:105) (cid:104) E (cid:105) ratio (cid:104) E (cid:105) (cid:104) E (cid:105) ratio (cid:104) E (cid:105) (cid:104) E (cid:105) ratio [MeV] ratio ratio [MeV] ratio ratio [MeV] ratiouranium-235 1 4.48 1 1 4.28 1 1 4.25 1uranium-238 1.53 4.59 1.024 1.56 4.45 1.040plutonium-239 0.64 4.26 0.950 0.65 4.13 0.965 0.66 4.04 0.951plutonium-241 0.93 4.47 0.998 0.90 4.23 0.988 0.91 4.13 0.971 tions become very similar. In other words, the difference in neutrino yield and mean energybetween the fissile isotopes is consistently predicted by the various flux models – which isnot surprising given that these differences have their origin in the fission yields.In practice, the current errors of any flux model are significant and a set of calibrationmeasurements at reactors of known fissile content is likely required to mitigate the effectof these uncertainties, particularly in view of the reactor antineutrino anomaly. A proof ofconcept at a theoretical level for these calibrations has been performed. On the experimentalside, the Daya Bay collaboration has demonstrated the ability to cross-calibrate a set of 8neutrino detectors to within better than 0.5%. The connection between fission rates and mass inventory requires a more detailed look at thereactor physics inside the core; our ultimate goal is to infer mass inventories. For a neutronflux which is constant in time and space, the fission rate and mass of a given fissile isotopehave a simple linear relationship f I = φ n σ I m I , (5)
48. Huber, “Spectral antineutrino signatures and plutonium content of reactors.”49. F.P. An et al., “A side-by-side comparison of Daya Bay antineutrino detectors,”
Nucl.Instrum.Meth.
A685 (2012): 78–97, doi: . m I is the mass of isotope I , σ I is the energy averaged fission cross section and φ n isthe neutron flux. Throughout the evolution of the core, all factors on the right hand side ofequation 5 will change. Due to burn-up effects, the mass m I will change and the neutron fluxtypically will be adjusted to compensate for changes in reactivity while maintaining constantpower. The accumulation of fission fragments will change the neutron absorption, which,in turn, alters the neutron energy spectrum; the cross section σ I will evolve as well. Theevolution of the isotopic content can be described by a set of Bateman equations and neutrontransport methods can be used to recompute the relevant cross sections. We have performedevolution or burn-up calculations for several reactor types using the SCALE software suite. For the further discussion it is useful to introduce fission fractions (cid:122) I , which are defined by (cid:122) I = f I (cid:80) I f I with (cid:88) I (cid:122) I = 1 . (6)This definition has the advantage that the problem can be phrased independently of reactorpower. For illustration, the time evolution of the (cid:122) I for a graphite moderated, naturaluranium fueled reactor is given in the left hand panel of figure 1, where the fission fractionsare shown as a function of the burn-up. (cid:122) Pu241 is very close to zero in this type of reactor andtherefore is not visible in this figure. The fission rate in uranium-238 stays constant sincethe amount of uranium-238 in the reactor changes very little with time. There is a clearanti-correlation between the fission fractions in uranium-235 and plutonium-239. The anti-correlation is nearly exact as shown in the right hand panel of figure 1 and we will make useof this later. In this context, it turns out that the burn-up is a useful variable which allowsa summary of the reactor inventory with a single number. Burn-up measures the number offissions which have occurred per unit of fuel mass or, in other terms, the amount of energyextracted; the unit for burn-up is MWd/t. For example, 1 tonne of fuel producing 5 MW for1 day yields a burn-up of 5 MWd/t; the same burn-up would be obtained by 1 tonne of fuel
50. “SCALE,” .
100 200 300 400 500 6000.00.20.40.60.81.0 Burn (cid:45) up (cid:64) MWd (cid:144) t (cid:68) (cid:184) C, NU238U 239Pu235U (cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230) (cid:184)
U235 (cid:184) P u239 Figure 1: The left hand panel shows the evolution of the fission fractions in a graphite moderatednatural uranium fueled reactor as a function of burn-up. The right hand panel shows the anti-correlation of the fission fractions in uranium-235 and plutonium-239. running at 1 MW for 5 days. Neglecting radioactive decays, the isotopic composition of bothsamples would be identical since the total number of fissions which took place is the same.As a result, the reactor core evolution is, to a very high degree of accuracy, a function ofonly the burn-up. That is, details of the power history, like innage factors and shut downs,have only a minor impact on the reactivity and fission fractions. The amount of plutoniumproduced depends on the details of the reactor operations and so does the resulting neutrinosignal. Therefore, we need to have a reasonably accurate model of the reactor power history,which in turn serves as input for a detailed reactor physics calculation. Neutrino emissionis a result of radioactive decay of fission fragments and therefore, fuel of the same burn-upwill have, to very good approximation, the same isotopic composition and will produce thesame distribution and amount of neutrinos at a given power level. Therefore, our abilityto predict the neutrino emission over time relies on an accurate model of the burn-up as afunction of time.This burn-up calculation also allows for the study of the time evolving relation betweenfission fractions and mass inventory as given in equation 5. We find, to very good accuracy,this is a linear relationship and the time evolution of the proportionality constant φ n σ I ,17hroughout the fuel cycle, is very small. Using a fixed value for φ n σ Pu239 , throughout thereactor cycle, induces a root mean square error of 2% for plutonium mass determinations fora graphite moderated reactor and errors of similar size for the other reactor types consideredlater. This type of sensitivity study needs to be performed for each reactor type and design.Also, the actual values of φ n σ I have to be determined for each specific case. The difference in the spectral neutrino yield, of the four fissile isotopes, can be used todisentangle the contribution of each of those isotopes to the total neutrino flux. In order todo so, we set up a binned χ -analysis, where the event rate in each bin n i is given as n i = N (cid:88) I f I (cid:90) E i +∆ E/ E i − ∆ E/ dE σ ( E ) S I ( E ) , (7)where E i is central energy of bin i , ∆ E is the bin width and σ ( E ) is the IBD cross section. N is an overall normalization constant set by the detector mass, or number of free protons,detection efficiency, and time interval of data taking. In order to compute the event rates n i ,we have to specify the four fission rates f = ( f U235 , f
U238 , f
Pu239 , f
Pu241 ). We denote the trueor input values for our calculation by a superscript 0, i.e. the true fission rates are f and,in the same way, we will denote the n i computed for the true values f as n i . We define the χ -function as χ ( f ) := (cid:88) i ( n i ( f ) − n i ) n i , (8)This χ -function will be zero for f = f . The allowed region for f is obtained by requiringthat χ ( f ) ≤ χ c , (9)where the critical value χ c is determined from a χ probability distribution with, in this case,4 degrees of freedom. If we are only interested in the total number of fissions in plutonium18iven by f Pu = f Pu239 + f Pu241 , the following marginalized function has to be used¯ χ ( f Pu ) = min f U235 ,f U238 ,κ χ ( f U235 , f
U238 , (1 − κ ) f Pu , κf Pu ) , (10)and in this case, since we are interested only in the single parameter f Pu the number ofdegrees of freedom is 1. Similarly, we can define a corresponding single parameter functionfor the measurement of reactor power.To relate a measured value of f Pu to the mass inventory, a reactor physics simulation isrequired. f Pu will be proportional to the plutonium mass, m Pu , in the reactor γ = m Pu f Pu , (11)where γ is the proportionality constant. Therefore, a measurement of f Pu translates into adetermination of m Pu . γ , in turn, depends on the details of the reactor physics as well as theinstantaneous reactor thermal power; note that according to equation 5, γ = 1 / ( φ n σ Pu ) andthus is inverse to the neutron flux density φ n . The determination of m Pu and its connectionto γ is clearly illustrated in figure 2, where we show the accuracy in the determinationof m Pu for a variety of reactor types as a function of the thermal power. This figure isbased on a full calculation of the reactor burn-up, where “C, NU” corresponds to a graphitemoderated reactor running on natural uranium and the dot on this line is the 5 MW e reactor,whose simulation details are explained in appendix A. “H O, HEU” and “H O, HEU + NU”correspond to the IRT with drivers only and to the IRT with drivers and targets, respectively.The details of the simulation are explained in appendix. C. The case “H O, LEU” is computedfor a typical pressurized light water reactor. We have taken a power history from one suchreactor, with a total fuel load of 72.4 MTU enriched to 3.7%. The case “D O, NU” describesa heavy water moderated reactor running on natural uranium modeled on a CANDU design
51. MTU stands for metric tonne of uranium and is often synonymous with metric tonnes of heavy metal(HM), where heavy refers to all actinides including plutonium. (cid:230)(cid:230) (cid:230) (cid:230)(cid:230) (cid:230)(cid:230) (cid:230) (cid:230) IAEA goal10 (cid:45) Thermal power (cid:64) MW (cid:68) Σ P ua cc u r a cy (cid:64) k g (cid:68) C , N U D O , N U H O , H E U H O , H E U (cid:43) N U H O , L E U Figure 2: Absolute accuracy in the determination of the plutonium content based on themeasurement of the neutrino spectrum as a function of the thermal power of the reactor. Thedifferent lines stand for different types of reactors as indicated by the labels: the first termindicates the type of moderator, whereas the second part denotes the fuel type, natural uranium(NU), low enriched uranium (LEU) and highly enriched uranium (HEU). This figure assumes a5 t detector, a standoff of 15 m from the reactor core, and 90 days of data taking. The horizontalline labeled “IAEA goal” indicates the accuracy which corresponds to the detection of 8 kg ofplutonium at 90% confidence level. with a 8.6 MTU natural uranium fuel load and running at 40 MW th . The 40 MW th point onthis line resembles, in many aspects, the Iranian reactor at Arak and the accuracy wouldbe at the level of 2.7 kg within 90 days. The details of the calculations for the “H O, LEU”and “D O, NU” can be found in appendix B. The horizontal line corresponds to a sensitivityto 8 kg plutonium within 90 days, which is the stated IAEA goal .For all of those quite different reactor types the accuracy of a m Pu measurement can bedescribed by the following simple relation δm Pu = 1 .
942 kg (cid:18) γ − kg s (cid:19) (cid:18) L m (cid:19) (cid:18) P th MW (cid:19) / (cid:18) tonnes M (cid:19) / (cid:18) days t (cid:19) / , (12)
52. David Albright and Christina Walrond,
Update on the Arak Reactor , technical report (Institute forScience and International Security (ISIS), 2013).53. The IAEA uses this interpretation of the term significant quantity as the design basis for planningroutine inspections in declared facilities. Any amount of fissile material diversion, or undeclared production,would be sufficient to warrant suspicion and follow-up activities to determine whether or not non-compliancemight be considered by the IAEA Board of Governors. L is the standoff of the neutrino detector, P th is the average thermal reactor power, M is the detector mass in tonnes (assuming 8 . × protons per tonne), and t is the length ofthe data taking period. Table II lists the corresponding values of γ , and using those values, Table II: The values of γ for a number of reactor types.reactor type C, NU H O, HEU H O, HEU+NU D O, NU H O, LEU γ [10 kg s ] equation 12 reproduces the results of the full calculation within a few percent. For graphitemoderated reactors, we find that the resulting δm Pu is significantly larger, by a factor ofat least 8.5, than for any other reactor type we have investigated. As we will show in thefollowing, the fact that neutrino safeguards still yield meaningful results and are applicablefor this reactor type is a testimony to the great versatility and power of this technique.For most reactor running conditions, the variation in γ is very small and depends onlyvery weakly, at the level of a few percent, on burn-up and reactor history. This implies thatour result most likely will hold up even for detailed 3-dimensional reactor physics calculations,taking into account spatial burn-up variations.We further observe, that for reactors with a thermal power in excess of 1 GW th , which isthe bulk of all reactors globally used for electricity production, this approach to safeguardswill have difficulties in meeting the IAEA goal of detection of 1 significant quantity, whichfor plutonium is 8 kg, within 90 days . On the other hand, neutrino safeguards is quitestraightforward for research, small modular reactors, and plutonium production reactors.As discussed in the previous section, the fission fractions and thus the fission rates are not independent from each other but are coupled by the physics inside the reactor; for anillustration, see the right hand panel of figure 1. In trying to determine the plutonium massinventory, we can make use of these correlations. Basically, reactor physics determines howthe fission rates evolve together with burn-up. Therefore, a reactor model will provide the
54. Plutonium in irradiated fuel is a so-called indirect use nuclear material and the precise IAEA goal is a90% or higher confidence level detection of the diversion of 1 significant quantity within 90 days, accordingto International Atomic Energy Agency,
IAEA safeguards glossary (International Atomic Energy Agency,2002).
The DPRK is rather unique in many regards, including its use of a nuclear weapons programas a bargaining tool. The direct tactical use of its small and crude nuclear arsenal against theU.S., or its regional allies like South Korea or Japan, is presumably deterred by the threat ofU.S. retaliation. It is a serious concern that North Korea may share its nuclear know-how,materials, or even a fully functional weapon with third parties, but the fear of the likelyattribution in case of a nuclear incident and the accompanying U.S. reaction may counteractthis risk. So far, North Korea obtained the largest benefit from its nuclear adventures byoffering to abstain in the future. For the use as a bargaining tool, it is desirable to create alarge degree of ambiguity about the type and scope of nuclear activities. At the same time,
55. Siegfried S. Hecker and William Liou, “Dangerous Dealings: North Korea’s Nuclear Capabilities andthe Threat of Export to Iran,”
Arms Control Today (2007). Thethree nuclear tests in 2006, 2009, and 2013 have removed a great deal of ambiguity aboutthe kind and goal of North Korea’s nuclear activities, but they shed no light on the scope ofactivities and the size of the resulting arsenal. New concerns have surfaced relating to theuranium enrichment program. The DPRK signed the NPT on December 12, 1985; a safeguards agreement entered intoforce on April 10, 1994; and notice of withdrawal from the treaty was given on January 10,2003. On February 26, 1993, the IAEA called for special inspections, which in retrospectmay have been counterproductive, to resolve the discrepancies found during the first safe-guards inspections in 1992. The issue of contention was the amount of plutonium the DPRKhad separated from spent nuclear fuel – North Korea declared it produced about 90 g, butIAEA data allowed for the possibility of a much larger amount, maybe as much as 14 kg, which would be sufficient to build two or more nuclear bombs. On March 12, 1993, theDPRK declared its intention to leave the NPT by June 12, 1993 after being threatened withspecial inspections but was persuaded by the U.S. on June 11 not to do so. A detailedrepresentation of the time line is given in figure 3.IAEA safeguards ended in 2003 and therefore we would like to focus on the time before2003. Would antineutrino reactor safeguards have been able to reduce the pre-2003 ambigui-ties about the DPRK’s plutonium production program and what would the potential impactof this information on the development of the crisis have been?
56. Mohamed ElBaradei,
The Age of Deception: Nuclear Diplomacy in Treacherous Times (MetropolitanBooks, 2011).57. Siegfried Hecker, “A Return Trip to North Korea’s Yongbyon Nuclear Complex” (Center for Interna-tional Security and Cooperation, Stanford University, 2010).58. David Fischer,
History of the International Atomic Energy Agency : the first forty years (The Agency,1997).59. ElBaradei,
Age of Deception .60. Don Oberdorfer,
The two Koreas: a contemporary history (Basic Books, 2001), p. 269.61. David Albright, “How much plutonium does North Korea have?”
Bulletin of the Atomic Scientists P R K c on c l ude ss a f egua r d s ag r ee m en tf o r t he I R T D P R K w i t hd r a w s f r o m N P T M W e r ea c t o r s hu t do w n f o r ys J an1977 J an2003 M a r S a f egua r d s ag r ee m en t en t e r s i n t o f o r c e F i r s tI AEA v i s i t F i r s tI AEA i n s pe c t i on S e c ond I AEA i n s pe c t i on T h i r d I AEA i n s pe c t i on F o r t h I AEA i n s pe c t i on D P R K de c li ne s a f i ft h i n s pe c t i on S pe c i a li n s pe c t i on r eque s t ed10 A p r M a y M a y J u l S ep1992 02 N o v N o v F eb1993 5 M W e r ea c t o r s hu t do w n 5 M W e r ea c t o r de f ue li ng s t a r t s D P R K w i t hd r a w s f r o m I AEA M W ede f ue li ng c o m p l e t e A g r eed F r a m e w o r k i ss i gned A p r M a y J un1994 15 J u l O c t D P R K be c o m e s pa r t y t o t he N P T S t a r t o f M W e r ea c t o r D e c J an1986 Figure 3: Time line of events e reactor at Yongbyon in 1986.In 1989 there was a 70 day shutdown, providing an opportunity to unload between 50-100%of the spent fuel in the core. In its initial declaration to IAEA in 1992, North Korea indicatedthat they ran a one-time reprocessing campaign in 1990 that resulted in 90 g of plutoniumfrom a limited number of damaged fuel rods removed during the 1989 shutdown. The resultsof IAEA environmental sampling conducted during the first safeguards inspection in 1992,however, indicated at least three campaigns of reprocessing in 1989, 1990, and 1991 whichin turn admits the hypothesis that a significant fraction of the spent fuel had been removed in1989 and subsequently reprocessed. As a result, a larger amount of separated plutonium mayhave been obtained by the DPRK, possibly sufficiently large to build two or more nuclearbombs. Given the ramifications of these findings, IAEA Director General Hans Blix insistedon a definitive resolution of this question as a precondition to declare the DPRK to be incompliance with its commitments under the NPT. In particular, finding and sampling thereprocessing waste streams was a priority for IAEA, eventually triggering the request forspecial inspections. The diplomatic exchange between IAEA and the DPRK dragged on inparallel with negotiations between the DPRK and the U.S.; the latter eventually leading tothe Agreed Framework. In April 1994 North Korea forced the issue by beginning to unloadspent fuel from the reactor core. An analysis of the gamma-radiation of spent fuel taken atknown positions in the reactor core would have resolved the question of how much spent fuelwas discharged in 1989 and whether the North Korean declaration was correct. Knowing theoriginal position of a sample in the reactor core is crucial for this analysis, since the fissionrate is higher in the center than at the edge of the core; for technical details of this methodsee section 6.1. However, the unloading proceeded very fast and it appears as if the operatorstook deliberate steps to obliterate any information about the original position of each fuelelement in the reactor; effectively, IAEA inspectors could only observe the unloading but
62. Albright, “How much plutonium does North Korea have?”63. O. Heinonen, Interview by PH, April 16 2013. igure 4: A map of relevant boundaries and geographies of the Yongbyon nuclear facility. Contoursshow expected inverse beta-decay event rates for a 5 t detector over the course of a year. X’smark the location of various neutrino detectors used in the paper. The satellite image on whichthis map is based was taken on May 16, 2013 by GeoEye-1. were unable to take any meaningful measurements of any individual fuel elements as theywere being removed, or to make a connection between the fuel elements and the core locationsthey had occupied. As a result, crucial evidence was denied to the IAEA and on June 2, 1992Blix declared that the ability to resolve the issue had been “seriously eroded”. The fueldischarged in 1994 was canned using U.S. equipment and subsequently was put into storageand was under IAEA surveillance until 2003, when the DPRK declared its withdrawal fromthe NPT. The 1994 crisis was resolved by the so called
Agreed Framework under which theDPRK halted any plutonium production and fuel reprocessing in exchange for the promise toobtain two pressurized light-water reactors at not cost. The
Agreed Framework unraveledin 2003 and eventually, in 2006, North Korea conducted its first nuclear test explosion.
A comprehensive account of the history of the North Korean nuclear program is provided byHecker. For our purposes, three facilities are relevant: a Soviet supplied research reactor
64. Albright, “How much plutonium does North Korea have?”65. Wit, Poneman, and Gallucci,
Going Critical .66. Hecker and Liou, “Dangerous Dealings”; Siegfried Hecker, “Lessons learned from the North Koreannuclear crises,”
Dædalus
Winter (2010): 44–56. th , the IRT; a graphite moderated reactor with a thermal powerof approximately 20 MW th , which generally is referred to by its electrical power, hence thename 5 MW e ; and the Radiochemical Laboratory, which is a reprocessing facility whichallows for the extraction of plutonium from the spent fuel from the 5 MW e reactor. Thesefacilities and their relative locations are shown in figure 4. Features such as the river andrelevant buildings are outlined, neutrino detector locations are marked, and IBD event rateiso-contours are shown.In the 1960s, the IRT was supplied by the Soviet Union. This reactor is a light-watermoderated reactor running on highly enriched uranium, with enrichment from 10% to 80% .The Soviet Union also provided the HEU fuel until its own demise in the 1990s. Withthis reactor, the Isotope Production Laboratory, a facility for handling irradiated materials,was provided. The nominal power of this reactor is 8 MW th . Using the laboratory, early,small scale plutonium separation experiments may have been conducted with fuel or targetsirradiated in this reactor. North Korea started serious fuel cycle activities in the 1980s and the plan was to build andoperate three gas-cooled, graphite moderated, natural uranium fueled reactors. A 5 MW e and 50 MW e reactor were foreseen for the Yongbyon site and a 200 MW e power reactor wasplaned at Taechon. The design followed the British Magnox design, where Magnox is thename of the alloy used for the fuel cladding: magnesium non-oxidizing. The thermal power ofMagnox reactors is typically 4-6 times higher than the above quoted electrical power, so theyare much less efficient than, for instance, pressurized light-water reactors, where this factoris closer to 3. Apart from efficiency, the choice of Magnox has another severe drawback:Magnox fuel cladding corrodes in contact with water such that long term storage of spentfuel under water is not possible. This makes encapsulation or some level of reprocessing
67. Hecker, “Lessons learned.”68. David Albright and Kevin O’Neill, eds.,
Solving the North Korean Nuclear Puzzle (ISIS Press, 2000),p. 148.69. Ibid., p. 92. The attractive features of this design are its simplicity and that itdoes not require uranium enrichment or the use of exotic moderators like heavy water. So,this reactor type was well adapted to North Korean indigenous industrial capabilities. Atthe same time, Magnox reactors were originally designed as dual-use facilities to produceboth electricity and weapons-grade plutonium.The amount of plutonium produced in a reactor can be estimated if the integrated neutronflux, which is proportional to the total energy produced, is known, or equivalently if a complete history of the reactor power is available. It turns out that all uncertainty aboutthe produced amounts of plutonium center on the issue of the completeness, and to a lesserdegree, the uncertainty of the record of the power history. To obtain the produced plutoniumin usable form, the reactor has to be shut down , the irradiated fuel rods removed, and theplutonium then needs to be chemically separated from the spent fuel at the RadiochemicalLaboratory. The location of the various facilities can be seen in figure 4.The time evolution of the burn-up for the 5 MW e is shown in figure 5 which has beenadapted from Nuclear Puzzle and is deemed accurate. The information in this figure isthe backbone of the analysis presented here and our quantitative results are based on thisinformation. The blue curve is based on the declarations made by the DPRK and, thus, theassumption is that no major refueling has taken place in 1989. The orange curve is derivedassuming that the full core has been replaced with fresh fuel in 1989 under the constraintof arriving at the same final burn-up. These numbers can be readily converted into reactorthermal power levels using the fact that there are approximately 50 tonnes of uranium inthis reactor. The power levels then form the input for a detailed calculation of the reactorisotopic composition and fission rates for the various fissile isotopes. The software we used
70. Albright, “How much plutonium does North Korea have?”71. In principle, Magnox reactors can be refueled under load, but the DPRK seems not to have masteredthis technology at that time.72. O. Heinonen, Interview by PH, April 16 2013, Heinonen confirms, looking at fig. VI.2 of
Nuclear Puzzle that this is an accurate description of the burn-up.73. Albright and O’Neill,
Nuclear Puzzle .
500 1000 1500 2000 2500 30000100200300400500600700 70d Shutdown 1st Inspection '94 ShutdownTime since Jan 1986 (cid:64) d (cid:68) B u r n (cid:45) up (cid:64) M W d (cid:144) t (cid:68) Case 1 (cid:72) no diversion (cid:76)
Case 2 (cid:72) full core exchange (cid:76)
Figure 5: Burn-up of the fuel in the 5 MW e reactor as function of time measured in days sinceJanuary 1, 1986. The blue curve is based on the values declared by the DPRK, i.e. no majorrefueling has taken place in 1989. The orange curve is derived assuming that the full core hasbeen replaced with fresh fuel in 1989. Figure adapted from Albright and O’Neill, Nuclear Puzzle . to compute the relevant reactor parameters is called SCALE and is considered a standardmethod for this type of problem. The details of the calculation can be found in appendix A. The basic analysis techniques developed in the previous section can now be applied to thespecific situation in the DPRK in the time frame of 1986-1994. The central question forthe international community, after the initial discrepancies appeared in 1992, was how muchplutonium the DPRK had separated. The lower bound on this quantity is represented byassuming that the DPRK’s initial declaration to IAEA was quantitatively correct, i.e. only90 g of plutonium were separated from a few hundred damaged fuel elements discharged andreplaced during the 1989 shutdown. The upper bound on the amount of separated plutoniumis obtained by assuming that the full core with a burn-up of approximately 200 MWd/t wasdischarged in 1989, containing 8.8 kg of plutonium and that this full core was subsequently
74. “SCALE.” Other authors estimate thetheoretical upper limit of the amount of plutonium to be as large as 4 kg.As far as the 5 MW e reactor is concerned, at the time of the first IAEA inspection in1992, the burn-up and reactor power were the same for both the extreme cases (see figure 5).Therefore, our analysis will include the hypothetical scenario where neutrino safeguards wereapplied before and after the 1989 shutdown . The specific unique capability represented byneutrino safeguards in this case derives from the ability to measure the power history andburn-up independently – any mismatch indicates a fuel diversion.In the PUREX process for reprocessing, the fission fragments remain in the aqueous phaseand therefore will end up in the waste. Some of these fission fragments produce neutrinosabove IBD threshold even after a considerable time interval has elapsed, which we will refer toas long-lived isotopes (LLI), in particular: strontium-90 with a half-life of 28.9 y, ruthenium-106 with a half-life of 372 d, and cerium-144 with a half-life of 285 d. These three isotopeshave large direct fission yields and are produced in amounts which are proportional to thenumber of total fissions and thus are accurate tracers of burn-up. Detecting neutrinos fromLLI is a direct method to find reprocessing wastes and, in principle, also yields an estimateof the amount of plutonium separated. Given the high penetrating power of neutrinos, thismethod is equally applicable to buried wastes.Finally, neutrinos can travel arbitrary distances, and thus a neutrino detector deployed
75. O. Heinonen, Interview by PH, April 16 2013, Heinonen estimates that the upper limit is between0.5-1 kg, based on the detailed fuel burn-up data the IAEA obtained as part of its safeguards agreement forthe IRT.76. Albright and O’Neill,
Nuclear Puzzle , p. 120.77. In principle, the amounts of the long-lived isotopes strontium-90, ruthenium-106, and cerium-144 will bedifferent between the two irradiation histories which leads to differences in the low energy neutrino spectrumbelow 3.6 MeV. However, extensive calculations show that the resulting event rate differences in 1992 aretoo small to be reliably detected.
500 1000 1500 2000 2500 300005101520 70d Shutdown 1st Inspection '94 ShutdownTime since Jan 1986 (cid:64) d (cid:68) T he r m a l po w e r (cid:64) M W (cid:68) (cid:64) d (cid:68) B u r n (cid:45) up (cid:64) M W d (cid:144) t (cid:68) Figure 6: In the left hand panel, 1 σ sensitivities to reactor power are shown for varying datacollection periods using a 5 t detector at 20 m standoff from the 5 MW e reactor. Fission fractionsare free parameters in the fit. In the right hand panel, 1 σ sensitivities to burn-up are shown,where power is a free parameter in the fit. The blue curve shows the history under the assumptionof no diversion. The orange curve shows history for the case of a full core discharge in 1989. for safeguarding the IRT would also be sensitive to neutrinos from the 5 MW e , especiallyduring times when the IRT is shut down. This signal will allow a remote power measurementwhich can distinguish the two cases shown in figure 5. e reactor In the following analysis, sensitivities to power, burn-up, and plutonium content are deter-mined based on the declared power history. This history is displayed as blue curves in thevarious figures in this section. Comparisons are made to a hypothetical undeclared coreswap to a fresh reactor core during the 70 day shutdown period, displayed as orange curves.The difficulty in determining the difference between the two curves lies in the fact that after1992, power and burn-up are the same. As seen in figure 11, after the 1st inspection, all thefission rates from the four primary fissioning isotopes are identical with or without diversion.For the following analyses, a standard 5 t detector at 20 m standoff from the reactor is used,which for a data taking period of one year corresponds to about 95,000 events.A power sensitivity computation is first considered. The analysis is done using the31ollowing χ -function χ = (cid:88) i n i . (cid:34)(cid:32) N P th (cid:88) I (cid:122) I S I,i (cid:33) − n i (cid:35) , (13)where (cid:122) I is the fission fraction for isotope I , n i is the measured number of neutrino eventsin energy bin i , and S I,i is the neutrino yield in energy bin i for isotope I . P th is the thermalpower and N is a normalization constant. Moreover, the fission fractions (cid:122) I are subject toa normalization constraint as given in equation 6.The resulting 1 σ sensitivities are shown in the left hand panel of figure 6. This analysisassumes precise knowledge of the distance from the reactor to the detector and treats themboth as points. Any uncertainty in the geometric acceptance will directly relate into anuncertainty of the normalization constant, N , and thus into an uncertainty in the power P th .Neglecting this potential source of systematic uncertainty, a power accuracy of around 2%can be achieved.A similar analysis can be done to determine the sensitivities for burn-up, BU , usingequation 13. In this circumstance, P th is free in the fit and the fission fractions (cid:122) I are nowfunctions of burn-up, determined by a reactor core simulation as described in appendix A.The results of this analysis are shown in the left hand panel of figure 6. Burn-up acrossthe history of the reactor has an error of ∼
100 MWd/t. Closely related to the burn-up is the amount of plutonium in the nuclear reactor. This analysis is done again usingequation 13. This time, P th as well as (cid:122) U235 and (cid:122)
U238 are free parameters as well as therelative contribution of the two plutonium fission rates, κ , and the resulting sensitivities areshown as dashed black lines in figure 7. Alternatively, one can use the burn-up sensitivity toconstrain the plutonium content as well. After computing burn-up errors, a reactor modelis used to compute the change in plutonium fissions. This is shown as the solid black errorbars. These errors are given both in terms of raw plutonium fissions in the left hand panelas well as the corresponding plutonium masses in the right hand panel. In the right hand32
500 1000 1500 2000 2500 3000024681012 70d Shutdown 1st Inspection '94 ShutdownTime since Jan 1986 (cid:64) d (cid:68) P u f i ss i on s (cid:64) s (cid:45) (cid:68) Pu fission variationBurn (cid:45) up variation 0 500 1000 1500 2000 2500 3000010203040 70d Shutdown 1st Inspection '94 ShutdownTime since Jan 1986 (cid:64) d (cid:68) F i ss il e P u (cid:64) k g (cid:68) Free neutron exclusion Pu fission variationBurn (cid:45) up variation
Figure 7: 1 σ sensitivities to plutonium are shown for varying data collection periods using a5 t detector at 20 m standoff from the 5 MW e reactor. The blue curve shows the plutonium-239history under the assumption of no diversion. The orange curve shows the plutonium-239 historyif there had been diversion. Black dashed error bars show the 1 σ sensitivity by measuring theplutonium fission rates with uranium fission rates and reactor power free in the fit. Solid blackerror bars show the 1 σ sensitivity determined by constraining the burn-up using a reactor model.The left plot shows the errors on absolute plutonium fission rates and the right plot show thecorresponding errors for plutonium mass with a shaded exclusion region from the assumption thatall neutrons not needed for fission are available for the production of plutonium. panel, a very naive exclusion region is shown for comparison. It assumes that each of the1.7 neutrons per fission not being used to sustain the chain reaction is instead available toproduce more plutonium. This limit is shown as the shaded region in the right hand plot. The IRT is assumed to run for a 250 day period followed by a 100 day shutdown, andthe fission rates are computed in appendix C and shown in figure 12. The natural uraniumtargets provide much more uranium-238, changing the fission fractions substantially andallowing an order of magnitude increase in plutonium-239 production and fissions. As withthe 5 MW e reactor, it is assumed that a 5 t neutrino detector is placed 20 m away from thisreactor. The χ from equation 13 can be used to determine the thermal power to within0.6 MW in each 50 day period. All other things the same, the addition of targets will increasethe power output of the reactor. As long as the detector distance and mass were sufficiently
78. Albright and O’Neill,
Nuclear Puzzle , pp. 148-165.
50 100 150 200 2500100200300400500 0.2.55.7.610.1Time (cid:64) d (cid:68) F i ss il e P u (cid:64) g (cid:68) P u f i ss i on s (cid:64) s (cid:45) (cid:68) Pu fission variationBurn (cid:45) up variation 0 50 100 150 200 2500.00.51.01.52.02.53.03.5 0.1.63.96.28.410.7Time (cid:64) d (cid:68) F i ss il e P u (cid:64) k g (cid:68) P u f i ss i on s (cid:64) s (cid:45) (cid:68) Free neutron exclusion Pu fission variationBurn (cid:45) up variation
Figure 8: 1 σ sensitivities to reactor plutonium fissions are shown for 50 day collection periodsusing a 5 t detector, 20 m away from the IRT reactor. Black dashed error bars show the 1 σ sensitivity resulting from measuring the plutonium fission rate with with uranium contributionsand power free in the fit. The solid black error bars show the 1 σ sensitivity determined using aburn-up model. The left plot shows driver only results and the right plot shows results for driverand targets combined. well known, the errors would be small enough to clearly notice the power difference causedby the addition of breeding targets. At the same time, it would be trivial for the operatorto adjust the power with targets to remain the same as without targets, which would reduceplutonium production by about 25%.The 1 σ errors on plutonium content can be determined by measuring the fission ratesusing the χ prescription in equation 13 and then converting these to a plutonium massusing equation 5. Alternatively, one can determine the burn-up in conjunction with a reactormodel and then infer the errors on plutonium mass inventory. The results of the analysisare shown in figure 8. Similar error bars are found on raw plutonium fission rates, with andwithout the targets. Note, that these are also similar to the 5 MW e reactor results. Verydifferent, however, are the sensitivities to the mass of plutonium. In the case with onlydrivers, a neutrino detector would be sensitive to tens of grams of plutonium. With both thedrivers and targets, there is an order of magnitude increase in the errors into the hundredsof grams of plutonium. The difference is even more pronounced in comparison to the 5 MW e reactor, where plutonium mass sensitivities in the multi-kg range are obtained. Despitesimilar sensitivities for plutonium fission rates, the sensitivity to core inventory is strikingly34ifferent for the reasons explained in detail in section 2.4. The neutron flux density in thefuel containing the plutonium is very different for the two configurations of the IRT withand without the breeding targets. Since the change in plutonium fission rates is relativelysmall between these two configurations, we have to conclude that neutrino safeguards is noteffective in determining which configuration is used. Therefore, the spread in plutoniummass predictions between the two configurations has to be taken as error, which is 0.36 kg,over one 250 day run. Taking the upper end of the range of plutonium produced in the IRT of 4 kg, we see that this requires about 8-10 reactor cycles. Since the errors from a neutrinomeasurement between each cycle are statistically independent we find the total error from aneutrino measurement taking 8 cycles to be 0 .
36 kg √ . e reactor power measurement at IRT An additional benefit of having a neutrino detector at the IRT reactor is that it would also besensitive to neutrinos from the 5 MW e reactor. This is particularly useful during times whenthe IRT is shut down, which happens for approximately 100 days every year. This willyield two measurement periods of 100 days each for the reactor power of the 5 MW e reactorduring the crucial time, after the 70 d shutdown and before the first inspection, where thedeclared power was low, around 8 MW th , but would have been as high as 18 MW th , in orderto bring the second core to the same final burn-up, see figure 5.Data collection is assumed to start shortly after an IRT shutdown at a point whereall but the long-lived neutrino producing isotopes have decayed away, leaving only the LLI:strontium-90, ruthenium-106, and cerium-144. This occurs on the order of days. The numberof atoms for each of the LLI was computed using SCALE and is shown in table III. As in
79. Albright and O’Neill,
Nuclear Puzzle , p. 120.80. Ibid., pp. 148-149. (cid:64) MeV (cid:68) E v en t s LLI Contribution8 MW18 MW 1000 1200 1400 1600 1800 2000 2200 24000510152025 70d Shutdown 1st InspectionTime since Jan 1986 (cid:64) d (cid:68) T he r m a l po w e r (cid:64) M W (cid:68) Figure 9: In the left hand panel events are shown for 200 days of data collection 20 m from theshut down IRT reactor and 1.2 km from the running 5 MW e reactor. The IRT is assumed to onlycontribute to the detected neutrino spectrum through its long lived isotopes shown in black. The5 MW e reactor is assumed to be running either at the declared 8 MW th , as shown in blue, or at18 MW th , as shown in orange. The right hand panel shows the 1 σ sensitivities to reactor powerresulting from this measurement. The blue curve shows the power history under the assumptionof no diversion. The orange curve shows the power history if there had been diversion.Table III: Number of long-lived isotope atoms assumed shortly after IRT shutdown.Isotope strontium-90 ruthenium-106 cerium-144Amount (atoms) . × . × . × the previous sections, we use a 5 t detector at 20 m standoff from the IRT and 1.2 km fromthe 5 MW e reactor, see figure 4. Data is collected over two 100 day periods and the detectedspectrum is shown in the left hand panel of figure 9. The signal event numbers are smalland therefore we use the appropriate Poisson log-likelihood to define the χ -function χ = 2 (cid:88) i [ n i log n i n i − ( n i − n i )] with n i = N P th (cid:88) I (cid:122) I S I,i + LLI i , (14)where LLI i is the long lived isotope contribution in the bin i . Resulting sensitivities areshown in the right hand panel of figure 9. This corresponds to an uncertainty of about3.8 MW th during the periods of interest. The difference in reactor power for a second corewould be detected at 3.2 σ .This result implies that a larger detector could be used to safeguard several reactors36n a larger area. In particular, a detector that is sensitive to direction could identify thereactor that contributed the neutrino and get several power measurements simultaneously.Also, without the need to be close to a reactor, it could be placed underground allowing forgreater background reduction. In addition to directly monitoring reactors, neutrino detectors can be used for detection ofnuclear waste. With sufficient insight of where waste might be disposed, a nearby neutrinodetector can see the signature of LLI, even after years of storage. Table IV lists the numberof atoms of each of the three primary LLI that would be expected in the waste at the pointin time of the first inspection, roughly 3 years after the 70 day shutdown. In the followinganalysis, it is assumed that the complete core was removed during the 70 day shutdown andthe resulting reprocessing wastes are stored together in one of three locations: the “suspectedwaste site”, building 500, or the Radiochemical Laboratory. All three locations are shownin figure 4. For building 500, we assume that we can not deploy inside the hatched area,since this facility was declared to be a military installation exempt from safeguards access. The resulting standoff distances are shown in table V.
Table IV: Number of long-lived isotopes at day 2251 for a complete reactor core removed at day1156 and stored for 3 years.Isotope strontium-90 ruthenium-106 cerium-144Amount (atoms) . × . × . × Due to the low event statistics, a Poisson log-likelihood is used, as in equation 14, withthe difference that the reactor events from the 5 MW e are now background and the signalare the LLI i . Table V summarizes the results for each location. Figure 10 shows the event
81. Jocher et al., “Theoretical antineutrino detection, direction and ranging at long distances.”82. Albright and O’Neill,
Nuclear Puzzle ; O. Heinonen, Interview by PH, April 16 2013, Heinonen does notbelieve that the liquid, high-level waste was transferred to building 500.83. Albright and O’Neill,
Nuclear Puzzle , pp. 149-154. (cid:64) MeV (cid:68) E v en t s Reactor contributionLLI contributionTotal events
Figure 10: Total event rates are shown in purple for 1 year of integrated data collection starting in1992 with a 5 t detector 25 m from spent fuel and 1.83 km from the 5 MW e reactor. The reactorcontribution to total event rates are shown in red and long lived isotope contributions shown inblue. rate spectrum in the most promising of the setups considered, the case of the reprocessingplant. It is found that a detector around 25 m from the waste and 1.8 km from the 5 MW e reactor would have a 2 σ signal after 55 days of data collection. The strongest contributorto detection capability is the distance from the source. Table V: Events are integrated over 1 year with a 5 t detector. The waste corresponds to acomplete reactor core discharged in 1989 during the 70 day shutdown. Long lived isotopes aredecayed 3 years before the measurement starts. The expected time to achieve a 2 σ detection isgiven in the last column.Location Reactor Fuel Reactor Fuel χ σ Time [y]Distance [m] Distance [m] Events EventsBuilding 500 1980 80 10.1 0.9 0.34 ≥ ≥ In applying neutrino safeguards, like in conventional safeguards, we can use individual mea-surements taken at different times and apply them in combination to infer what actuallyhappened. The initial declaration to IAEA by the DPRK admits two extreme cases: both a38ery minor discharge of a few hundred fuel elements and a complete core discharge in 1989would yield the same overall core configuration at the time inspectors arrived in 1992, seefigure 5.The real strength of a neutrino detector is evident when it can measure over the historyof a reactor. As seen in figure 6, such a detector is capable of being very sensitive to reactorpower. Thus, if a neutrino detector was present for the lifetime of the reactor, the declaredpower would have to match the measured power at all times and, since the burn-up is just thetime integrated thermal reactor power, the burn-up could be inferred from a complete powerhistory. At the same time, a burn-up measurement, in contrast to an inferred burn-up value,can also be derived from a neutrino measurement, provided a reliable, but not necessarilyvery detailed or accurate, reactor model is available. As will be shown in section 6.1, giventhat the bulk quantities in terms of burn-up are the same between the two scenarios, allconventional methods which can address the issue of the second core also rely on a reactormodel. The diversion scenario that has been considered relies heavily upon the ability toadjust the power relative to the declared power so that both the power and burn-up matchat a later time. In the presence of a neutrino detector, the difference in burn-up will befrozen between the declared burn-up and the actual burn-up of the new core. The fact thata neutrino detector can simultaneously measure power, as well as fission fractions, is whatallows it to detect this difference in burn-up. To determine sensitivity to such a situation, amodified version of equation 13 is used χ = (cid:88) t (cid:88) i n i,t (cid:34) (1 + α detector ) P t th (cid:88) I (cid:122) I ( BU t ) S I,i − n i,t (cid:35) + (cid:18) α detector σ detector (cid:19) . (15)where t is indexing the time interval for which a measurement is available. α detector is adetector normalization parameter with uncertainty σ detector . P t th is the average reactor powerin each time bin t . (cid:122) I are the fission fractions which are a function of the burn-up in each39ime bin t , BU t . The burn-up as a function of time is given by BU t = (cid:32) t − (cid:88) τ =1 P τ th ∆ τM core (cid:33) + BU (16)where ∆ τ is the width of the time bin, BU the initial burn-up at the start of data takingand M core the mass of the reactor core in terms of fuel loading. If this initial burn-up BU is well known, as it would be if data collection began at start-up, such an analysis greatlyreduces the uncertainty in the total plutonium budget. In table VI, the total error budgetis given through the use of this method, labeled “method 2”, and is shown compared to theresults if only the burn-up but not the power history is measured based on the results of theprevious sections, labeled “method 1”. For method 2 we assumed that reactors start with awell known composition, that is BU = 0 and a detector related uncertainty σ detector = 1%is achievable and all the P t th are free parameters in the fit. In the case of the 5 MW e reactor,for both analyses, the question is what is the maximum change in BU x during the 70 dayshutdown. The value of BU x is translated into the resulting plutonium mass sensitivity byusing the reactor model. It is clear that method 1 is less accurate but does not rely oncontinuity of knowledge whereas method 2 is much more accurate but requires continuity ofknowledge. Method 2 still offers a significant advantage compared to conventional methodsby providing its results in real-time and not only at some later, unspecified time in the future.For completeness we also list the plutonium mass sensitivities from the indirect methodand the detection of reprocessing wastes in table VII. So far in this analysis, we have neglected backgrounds not related to neutrino emissions.The main backgrounds in inverse beta-decay detectors are: accidentals, where two uncor-related events caused by ambient radiation in the detector accidentally fulfill the delayedcoincidence requirements in both time and energy; fast neutron induced backgrounds, where40 able VI: Plutonium content and 1 σ uncertainties are given for two analysis techniques for boththe IRT and 5 MW e reactors. Due to the inability to reliably detect the presence of targets inthe IRT reactor, they are assumed to be in the reactor. The detection capability is given foreach 250 day run of the IRT. The 5 MW e reactor plutonium error is a combination of removedplutonium that may have occurred during the 70 day shutdown and the final plutonium contentin the reactor at the 1994 shutdown. The quantities are independent if data is only taken afterthe 1st inspection and correlated if taken from start-up. The flat burn-up analysis adds a fixedburn-up to each time bin and the final plutonium error is the final plutonium difference betweenthe burn-up increased data and the expected data. The power constrained analysis assumesthe starting fuel composition is known and the burn-up is given by the integration of the powerwith an assumed 1 % detector normalization uncertainty. The plutonium error is the maximumplutonium difference attainable through power increases and fuel removal (in the case of the5 MW e reactor). Values are given for 1 σ sensitivities for maximizing the plutonium available forcore 1 or core 2 respectively. Parenthesis are for uncertainties in cores using only data fromthe respective section. Core 3 and core 4 are additional fuel loads that are irradiated in the5 MW e reactor post-1994 according to Albright and Walrond, North Korea’s Estimated Stocks ofPlutonium and Weapon-Grade Uranium and are added for completeness. ∗ Using uncertainty from Albright and O’Neill,
Nuclear Puzzle † These two numbers are anti-correlated with a correlation coefficient of -0.962.
Reactor Final Method 1, 1 σ Method 2, 1 σ Burn-up Pu Burn-up Pu Burn-up Pu[MWd/t] [kg] [MWd/t] [kg] [MWd/t] [kg]IRT/run with targets 3550 0.47 3520 0.47 39 0.015 MW e from Core 1 178 8.83 178 9.5 ∗ N/A1st inspection Core 2 648 27.7 95 3.295 MW e from Core 1 178 8.83 138 (83) 6.68 (3.76 † ) 43 (1.9) 2.12 (0.11)start-up Core 2 648 27.7 52 (66) 1.81 (2.30 † ) 6.7 (6.9) 0.23 (0.24)5 MW e Core 3 307 14.6 51 2.17 3.2 0.145 MW e Core 4 255 12.3 53 2.36 2.7 0.12 a fast neutron enters the detector without leaving trace and scatters off a proton, which thenis confused with the primary energy deposition of a positron, and subsequently the neutronthermalizes and captures like a genuine neutron from inverse beta-decay; β -n backgrounds,where interaction with cosmic ray muons produces a short-lived radioactive isotope whichdecays by beta-delayed neutron emission, which mimics a neutrino event. The rate of acci-dentals is determined by the rate of ambient radioactive decays. Fast neutrons are a resultof cosmic ray interactions in materials surrounding the detector and thus depend on therate of cosmic ray muons; the same is true for β -n backgrounds. Therefore, the measured41 able VII: 1 σ uncertainties on the discharged plutonium for core 1 for the IRT parasitic mea-surement and for the detection of high-level reprocessing waste.Core 1 burn-up [MWd/t] Core 1 Pu [kg]Parasitic measurement 51 2.55Waste measurement Suspected waste site 56 2.76Reprocessing plant 34 1.67 background rates due to those two sources have to be scaled from the underground location,where most current neutrino detectors are located, to the surface. Neutrino detectors arecommonly put underground precisely to reduce these two sources of backgrounds, since deepunderground the cosmic muon flux is strongly attenuated. The scaling of the number ofbackground events is not purely given by the muon flux, but also, to some degree, dependson the average muon energy, the scaling is given by R ∝ φ µ (cid:104) E µ (cid:105) α , (17)where φ µ is the muon flux and (cid:104) E µ (cid:105) is the average muon energy which, at the surface,are 127 m − s − and 4 GeV, respectively. α ranges from 0 . − . , we can scale to a surface deployeddetector and find 1 d − t − fast neutron events and 43 d − t − β -n events, where 1 tonne isassumed to have the composition of CH . These rates exceed the accidental rates by a largefactor and therefore we can neglect the accidental backgrounds. This scaling is tested againstseveral data sets from different experiments spanning a depth range from 850 −
120 mwe andthe scaling is found to be accurate within a factor of two. At very shallow depths of less
84. Y. Abe et al., “Direct Measurement of Backgrounds using Reactor-Off Data in Double Chooz,”
Phys.Rev.
D87 (2013): 011102, doi: .85. J. Beringer et al., “Review of Particle Physics (RPP),”
Phys.Rev.
D86 (2012): 010001, doi: .86. Overburden is commonly quoted in units of meter water equivalent (mwe), typically 1 m of rock/soilcorresponds to about 2-3 mwe.87. Abe et al., “Backgrounds.”
Table VIII: Noise to signal ratios for a surface deployed detector.Source Fast neutron suppression β -n suppression5MW e required rejection factor can be reduced significantly by providing a moderate overburden of10-20 mwe, which, in principle, can be engineered into the detector support structure.Fortunately, there is a significant on-going experimental effort in several countries toaddress the R&D for neutrino detectors with greatly improved background rejection. Theseinitiatives are motivated by the search for a new particle called a sterile neutrino throughthe use of neutrinos from reactors with detectors placed within meters of the reactor core.The close proximity to a reactor core results in a high-background environment which caninclude a significant flux of fast neutrons and high-energy gamma-rays from the reactor itself.Almost all reactor sites under consideration offer only very minimal overburden of 10 mweor less. Therefore, these experiments face essentially the same level of problems in termsof signal to noise conditions as safeguards detectors would under the conditions outlinedin this paper. Specifically, there are, to name but a few, the PROSPECT collaboration inthe U.S., the DANSS project and NEUTRINO-4 in Russia, the STEREO project in
88. Abazajian et al., “Light Sterile Neutrinos: A White Paper.”89. Z. Djurcic et al., “PROSPECT - A Precision Reactor Neutrino and Oscillation Spectrum Experimentat Very Short Baselines” (2013).90. I. Alekseev et al., “DANSSino: a pilot version of the DANSS neutrino detector” (2013).91. A.P. Serebrov et al., “On possibility of realization NEUTRINO-4 experiment on search for oscillationsof the reactor antineutrino into a sterile state” (2013).
The events in 1994 put a premium on understanding the actual history of the North Koreanplutonium program; a vivid interest in this problem remains in the aftermath. The actualtext of the Agreed Framework states [. . . ] before delivery of key nuclear components, the DPRK will come into fullcompliance with its safeguard agreement with IAEA (INFCIRC/403), [. . . ] withregard to verifying accuracy and completeness of the DPRK’s initial report on allnuclear material [. . . ]
Section IV, Paragraph 3In 1994, and even today, there has been a need to resolve the question of whether therewas significant reprocessing prior to 1992. It comes as no surprise that actual methods,relying on more conventional means, were devised. We will briefly review those conventionalmethods.The unloading of the 5 MW e core in June 1994 provided a crucial opportunity to acquiredata that would allow a determination of whether there was a partial or complete coreunloading in 1989. The DPRK, aware of this possibility, tried to prevent IAEA from gainingthis information by unloading the core very quickly and, by doing so, made it impossibleto infer the exact position of a fuel element inside the core. Unfortunately, there is littlepublished on the details of how a measurement would have proceeded and we have to rely oninterviews with experts. The basic concept of the method is to map out the three dimensionalburn-up distribution inside the reactor core. If the declaration by the DPRK, that only a44ew hundred damaged fuel elements were replaced in 1989 were true, then there should be adiscontinuity in the burn-up distribution at the position of the replaced fuel elements. Onthe other hand, if more than those few hundred were replaced then discontinuities would showup at many more locations. If the whole core was replaced a continuous distribution wouldemerge. Overall, there are about 8,000 fuel elements and the goal is to find a discrepancyconcerning as few as several hundred fuel elements. Therefore, a sizable sample of about 300fuel elements is required. There are two principal methods to determine the burn-up ofspent fuel: one is destructive sampling with subsequent isotopic analysis and the other is tomeasure the characteristic radioactivity emanating from a spent fuel element. Destructivesampling was (and is) exceedingly difficult in this context. The other possibility to measureburn-up relies on measuring gamma-emission from mostly cesium-137, which is a good proxyfor burn-up. According to an expert from Los Alamos National Laboratory who was closelyinvolved in the 1994 DPRK issue, this technique would provide burn-up errors below 5% ifgood quality calibration data existed. This method, in principle, has been calibrated onBritish Magnox fuels. Applying this type of measurement to several hundred of the spentfuel elements and knowing their location in the core presents a viable method to reconstructthe three dimensional burn-up distribution. None of the interviewed experts was willingto make a statement as to what level of precision, in terms of partial core reloads andextracted plutonium amounts, would have resulted. However, we can put a lower limit onthe achievable error by assuming that the overall systematic errors are less than 5% and theerrors for individual fuel elements are in the 1-5% range. Therefore, the overall accuracyvery roughly should be in 1-5% range.This estimate coincides with the accuracy of a method which is based not on the samplingof spent fuel but instead on sampling the graphite moderator in the reactor. The idea is
92. Heinonen,
Interview ; B. Reid and C. Gesh, Phone interview by PH, May 15 2013.93. Wit, Poneman, and Gallucci,
Going Critical , p. 170.94. Reid and Gesh,
Interview .95. H. Menlove, private communication.96. Ibid. all the neutrons produced throughout the history ofreactor operation. Since plutonium results from neutron capture on uranium-238, the totalamount of plutonium produced is strictly proportional to the number of all neutrons. Evenreactor grade graphite contains traces of other elements like boron or titanium and boththese elements have stable isotopes, specifically boron-11 and titanium-49, which result fromneutron capture. Therefore, the ratios boron-10/boron-11 and titanium-48/titanium-49 willdecrease with the total neutron fluence. This graphite isotope ratio method (GIRM) was firstproposed by Fetter in 1993 and subsequently developed in considerable detail at PacificNorthwest National Laboratory. In an actual application, samples from the graphite wouldbe taken at a few hundred strategically chosen points throughout the core and the isotoperatios would be determined by mass spectroscopy. This data then can be used to reconstructa three dimensional neutron fluence distribution which then, in turn, can be converted to thetotal amount of plutonium produced in the reactor through its entire lifetime. This methodwas experimentally verified at a British reactor with an accuracy in the 1-5% range. Thismethod is quite invasive and requires extensive cooperation between the national authoritiesand operator of a reactor in question, and the state or organization carrying out the testing.It has the advantage that it is tamper resistant and the historical record, in the form of thegraphite moderator of the 5 MW e reactor, is still available.We have reviewed two methods: one based on gamma-ray emission of spent fuel and theother on isotope ratios in the graphite moderator. Both methods rely on several hundredsamples collected across the reactor and a subsequent reconstruction of three dimensionaldistributions of either burn-up or neutron fluence. For the first method we can only providea rough estimate of accuracy, whereas for GIRM the errors have been experimentally deter-
97. Steve Fetter, “Nuclear Archeolgy: Verifying Declarations of fissile-material production,”
Science andGlobal Security
Trawsfynydd plutonium estimate , 13528, technical report (Pacific Northwest NationalLaboratory, 1997).99. Patrick Heasler et al., “Estimation procedures and error analysis for inferring the total plutonium (Pu)produced by a graphite-moderated reactor,”
Reliability Engineering and System Safety
91 (2006): 1406–1413. .2 Neutrinos
Based on the quantitative results and the time-line of events in 1994, see figure 3, thefollowing scenario may have been put into effect: • The IRT is under full neutrino safeguards with a dedicated 5 tonne detector from 1978on, which is located outside the IRT reactor building at the southern wall. • The 5 MW e is under full neutrino safeguards with a dedicated 5 tonne detector fromMay 1992 on, which is located outside the 5 MW e reactor building at the western wall. • A search for neutrino emissions from the reprocessing waste is initiated in November1992. Three 5 tonne detectors are deployed: one at the reprocessing plant; one at thesuspected waste site, located above the center of the waste site; and one at building500, located right outside the southern fence.This scenario is fully consistent with the actual safeguards access the IAEA had and, inparticular, all detector deployment locations reflect actual physical access. As a result, thedetectors at the IRT and 5 MW e have a standoff of 20 m, the detectors at the suspectedwaste site and reprocessing plant have a distance of 25 m, and the one at the building 500has a distance of 80 m.Furthermore, we assume that in 1989 the DPRK discharged the complete first core, whichseems to be corroborated by the declaration of the DPRK in 2008 that it possesses 30 kg ofplutonium. After reprocessing of the spent fuel, the waste was stored somewhere in thereprocessing plant.
Finally, we also assume that the burn-up declared by the DPRK in1992 is indeed correct.
This completely specifies the scenario.The first relevant piece of data would be obtained by the IRT detector during periodswhen the IRT is shut down, about 100 days out of each year. The neutrino signal stemming
North Korea’s Estimated Stocks of Plutonium and Weapon-Grade Uranium .101. Heinonen,
Interview .102. Albright and O’Neill,
Nuclear Puzzle ; Heinonen,
Interview . e is clearly detectable at this detector location and providesa measurement of reactor power. In 1989, this signal would have been recorded but wouldnot have raised any special concern, since the 5 MW e was not under safeguards at this time.At most, this data would have helped to corroborate U.S. government analyses of satelliteimagery to ascertain the operational history of the 5 MW e . However, soon after the DPRKhad submitted its initial declaration to the IAEA, in May 1992, this data would have resultedin a discrepancy which, in combination with the results from environmental sampling wouldhave led to the conclusion that a large amount of plutonium had been separated in 1989.This measurement, according to table VII, has a sensitivity which corresponds to 2.55 kgplutonium, or equivalently to a 8 . / .
55 = 3 . σ detection, meaning the IAEA would haveknown that a significant fraction of the first core had been discharged with a confidence of1 in 1 900. Taking 4 kg of plutonium as the quantity needed for a nuclear bomb, this resulttranslates into a 1 in 13 confidence that the DPRK has at least enough plutonium for onebomb.In November of 1993, after a year of data collection, the detectors at the suspected wastesite would not have found anything nor would the detector at building 500, the former resultproving that only a small amount of high-level waste could be present at the suspectedwaste site and the latter being insignificant since the distance to the waste is too large. Thedetector at the reprocessing plant would have shown the presence of high-level radioactivewaste, corresponding to a plutonium accuracy of 1.67 kg. That is, with a confidence of1 in 1,000,000, the presence of reprocessing waste would have been confirmed. Moreover,with a confidence now of 1 in 270, it would have been known that enough plutonium forone weapon was processed. Six months later, in May 1994, the 5 MW e detector would haveconfirmed the burn-up declaration of the DPRK with an accuracy of 15%. In combination,these results would have implied a 56% chance of there being enough plutonium for two or Interview ; Siegfried Hecker,
Report of Visit to the Democratic People’s Republic of NorthKorea (DPRK) Pyongyang and the Nuclear Center at Yongbyon, Feb. 12-16, 2008. , technical report (Centerfor International Security and Cooperation, 2008). and the DPRK still would havewanted to obtain maximal material and political gains for eventually accepting any U.S.demands. Both parties would have remained keen to avoid war on the Korean peninsula,since for the U.S. the number of casualties and financial burden would have appeared difficultto justify and a war would clearly be against the interest of its close ally South Korea. For theDPRK, or more specifically its leadership, a war would constitute the ultimate catastrophe and its scientists are knowledgeable andcompetent experts, which certainly would have the ability to understand all implications ofneutrino safeguards as outlined here.
Given that they made every effort to conceal thetrue history, they also would have tried to thwart neutrino safeguards. Neutrino safeguardsis not necessarily more difficult to thwart than other means, it just requires different counter
Interview .106. Heinonen,
Interview ; Reid and Gesh,
Interview . Neutrino reactor monitoring offers unique capabilities which seem to make this method – asproposed more than 30 years ago – a useful tool for safeguards. Also, neutrino detectors havebeen continually refined since the days of Cowan and Reines and can be considered a maturetechnology. Given the mechanisms of neutrino production and detection, neutrino safeguardsprovide bulk measurements of reactor core parameters like power or burn-up. This is to becontrasted with the current safeguards approach which largely relies on item accountancyand, in particular, neither power nor fuel burn-up are actually measured by the IAEA orverified through independent calculations for any reactor. The IAEA is apparently satisfiedthat the existing arrangements are adequate for commercial power reactors of the boilingand pressurized water types, especially as long as a once-through fuel cycle without repro-cessing is considered. As a result, it has been difficult to show neutrino safeguards wouldprovide a decisive advantage in comparison to more conventional techniques, especially sinceneutrino detectors are larger and more expensive than most equipment currently used bythe IAEA. The fact that, in the literature, a rather diverse set of results, in terms of theapplicability to specific safeguards issues, is found may have a further detrimental effecton the perception of neutrino safeguards. These widely varying results can be attributed,
52o a large degree, on differing assumptions about detector parameters and different level ofstatistical treatment. The choice of detector parameters often is inspired by the wish to beparticularly realistic or thrifty, which seems to be a classical case of what Donald Knuthcalls premature optimization.
Furthermore, many analyses are rate-based which leads toserious deficiencies in sensitivities since there is a pronounced degeneracy between reactorpower and fission fractions when spectral information is ignored .In this paper we have taken a different approach – with the North Korean nuclear crisisof 1994 we have identified a real-world scenario in which traditional safeguard techniquesultimately were unable to resolve the key questions and for which sufficient technical infor-mation is publicly available to perform a detailed analysis. Moreover, we base our detectorparameters on the overall acceptable size and weight of the entire detector system, which weenvisage to fit inside a standard 20 ft intermodal shipping container. We also assume thatthe detector will be able to operate at the surface. Together with the standard packagingthis will provide a great deal of flexibility in the choice of deployment locations. The price topay is that the detector has to have excellent background rejection, which seems to excludesingle volume liquid scintillator detectors and favors finely segmented solid detectors, see sec-tion 5.6. Detectors with these capabilities currently do not exist, but a basic science questionrelated to the possible existence of a new particle, a so-called sterile neutrino, has triggereda large number of experimental efforts to perform reactor neutrino experiments at a rangeof several meters from compact reactor cores. These experiments face enormous challengesfrom reactor-generated backgrounds and, therefore, have to solve the background rejectionissue. Many of the new designs do not rely on large-area photo-detectors, which are typicallyhand-made in small numbers and therefore are very expensive. New detector designs havethe potential to become more affordable in industrial production. Therefore, there is amplereason to assume that within a few years detectors with the required characteristics will be
Communications of the ACM
17, no. 12 (1974):667–673.110. cf. compare the results in () with the ones in () th power and at heavy-water moderated reactors producingless than 0.1 GW th power. These results suggest that the heavy-water reactor at Arak inIran with an estimated thermal power of 0.04 GW th could be an ideal target for neutrinosafeguards, and a detection limit of 4.4 kg plutonium within 90 days at 90% confidence levelseems possible. Graphite moderated reactors, on the other hand, are more difficult due to arelatively low power density.We also developed an analysis method based on the fact that the isotopic abundance ofthe various fissile isotopes as a function of burn-up is governed by reactor physics and conse-54uently these quantities are correlated in a well-defined manner, as explained in section 2.3.In this case, the problem can be rephrased in terms of reactor power and burn-up and thisimproves the sensitivity by roughly a factor of two. The reactor model required for this typeof analysis does not have to be extremely detailed or accurate, since only the gross burn-upevolution is required.For our purposes, the North Korean nuclear program consists of three pieces: the IRT, an8 MW th light-water research reactor supplied by the USSR, which is fueled with HEU and hasbeen under IAEA safeguards since 1977; the 5 MW e reactor, a 20 MW th graphite moderatednatural uranium fueled reactor; and the Radiochemical Laboratory, a reprocessing facilitywhich can extract plutonium from the spent fuel using the PUREX process. The centralquestion was whether the fuel in the 5 MW e reactor was the original core load or whetherthere was an earlier undeclared refueling during the shutdown in 1989. The discharged fuelwould have yielded about 8.8 kg of plutonium, sufficient for at least one nuclear bomb. Ourtechnical analysis is to a large degree based on the data presented in Nuclear Puzzle , which weuse as input for detailed reactor core simulations for both the IRT and the 5 MW e reactors.The North Korean declaration of the burn-up history of the 5 MW e is such that the coreconfiguration in terms of measurable quantities like burn-up and reactor power is virtuallyidentical for both the one-core and two-core scenarios. Therefore, safeguards techniques,both conventional and neutrino-based, have to resort to secondary observables. In the caseof neutrinos, secondary signatures focus on a measurement of reactor power by a neutrinodetector deployed to implement the safeguards agreement for the IRT, which has been inforce since 1977. This signal is visible only when the much closer and, hence, brighterneutrino source represented by the IRT is not in operation, which occurs for about 100 daysper year. This measurement would provide evidence for the presence of a second core witha confidence of 1 in 1 900 (3.5 σ ), see section 5.Another secondary signature, which can be exploited with neutrinos is the detectionof the presence of reprocessing wastes, which contain long-lived fission fragments, some of55hich emit detectable neutrinos. Historically, three sites have been suspected to be thepotential disposal locations: Building 500, a suspected waste site, and the RadiochemicalLaboratory. The map in figure 4 shows the relative locations. For the latter two sites aneutrino safeguards detector would have been able to detect the presence of the reprocessingwastes with a confidence of better than 1 in 600, see section 5. For the 1994 crisis, theapplication of neutrino safeguards could have resulted in significantly reduced uncertaintyabout North Korean intentions.In a more general context, we also studied the resulting sensitives assuming that neutrinosafeguards had been available from the start-up of the 5 MW e reactor and showed that if acontinuous measurement of reactor power by neutrinos had been available, which could thenbe compared to a measurement of the burn-up by neutrinos at a later point, there wouldhave been very little room for undeclared plutonium production or refuelings; accuraciescorresponding to 1-2 kg of plutonium would have been achieved, see table VI. Our workshows that even graphite moderated reactors can be safeguarded successfully using neutrinomonitoring. Furthermore, we compared the abilities of neutrino safeguards with conventionalcapabilities in section 6.1 and found that those conventional techniques have to rely on alevel of reactor physics modeling comparable to the more advanced analysis techniques wehave presented. Provided the extensive effort and funds required for such modeling canbe expended, the overall accuracy of conventional techniques should be in the 1-5% range,whereas neutrino techniques are in the 5-15% range in terms of plutonium content. Thecrucial advantage neutrino safeguards offer stems from the near real-time data acquisitionduring reactor operation, whereas the conventional methods require a reactor shutdownand a defueling of the reactor. Neutrino safeguards also is entirely non-invasive; at 20 mstandoff the detector can be deployed outside the reactor building. In the context of break-out scenarios, deferring the ability to know how much plutonium was produced or whether adiversion has taken place until a later point in time, when there is no guarantee of safeguardsaccess at the required level, is problematic – but this is what conventional techniques have56o rely on and have done in the case of the DPRK, with the known result.To summarize, using the North Korean nuclear crisis as a virtual laboratory, we havefound by detailed technical analysis that neutrino safeguards for water moderated reactorswith a thermal power less than 0.1-1 GW th can meet the IAEA detection goals in terms ofplutonium content and timeliness. This makes neutrinos a viable choice for many researchreactors, small, e.g. 40 MW th , plutonium production reactors, and for most of the plannedcommercial small modular reactors. Small modular reactors would allow for the inclusionof a neutrino safeguards system at the design stage. In the specific North Korean case,we find that neutrinos provide an accuracy which is marginally worse than conventionalmethods and, qualitatively, the difference in accuracy seems to be irrelevant. At the sametime, neutrinos allow conclusions about the plutonium content and potential diversion tobe drawn in close to real-time, whereas conventional methods provide the information onlyafter the fact, once the reactor is shut down and defueled. For all of these applications,neutrino detectors have to work with minimal or no overburden and the lower the residualbackground is, the more versatile the resulting system will be. For very low backgrounddetectors, remote power measurements and the detection of reprocessing wastes becomes anattractive possibility, in particular for purposes of nuclear archaeology. Acknowledgements
We thank R. Gallucci, C. Gesh, O. Heinonen, H. Menlove and B. Reid for their willingness tobe interviewed for this project. We also thank A. Erickson, L. Kalousis, J. Link, C. Marianiand in particular T. Shea for their expert opinions on many of the technical issues involvedregarding nuclear reactors, neutrino detectors, and safeguards. We thank M. Fallot forproviding reactor neutrino fluxes in machine readable format and we also acknowledge usefuldiscussions about solid, segmented neutrino detectors with A. Vacharet and A. Weber. Thiswork was supported by the U.S. Department of Energy under contract DE-SC0003915 and57y a Global Issues Initiative grant by the Institute for Society, Culture, and Environment atVirginia Tech. 58
500 1000 1500 2000 2500 300000.050.10.150.20.25 70d Shutdown 1 st Inspc. '94 Shutdown0 500 1000 1500 2000 2500 3000510152025 70d Shutdown 1 st Inspc. '94 Shutdown0 500 1000 1500 2000 2500 3000494504950049550496004965049700 70d Shutdown 1 st Inspc. '94 Shutdown0 500 1000 1500 2000 2500 3000320330340350360 70d Shutdown 1 st Inspc. '94 ShutdownAlbright0 Case ICase II U235U238Pu239Pu241Time after Jan. 1, 1986 (cid:64) days (cid:68) F i ss il eabundan c e (cid:64) k g (cid:68) U235Pu239U238Pu241Case IICase I0 500 1000 1500 2000 2500 300010
70d Shutdown 1 st Inspc. '94 ShutdownTime after Jan. 1, 1986 (cid:64) days (cid:68) F i ss i on s (cid:64) s e c (cid:45) (cid:68) Figure 11: Fissile content of the 5 MW e reactor (left hand panel) and the corresponding fissionrates (right hand panel) A The 5 MW e reactor The 5 MW e reactor is a graphite moderated and reflected Magnox reactor using naturaluranium fuel, based on the British Calder Hall design, and runs with a nominal powercapacity of 20-25 MW th . The average burn-up of the fuel elements is 635 MWd/t. The 5 MW e reactor contains 812 vertical channels, each with up to 10 fuel elements per channel. Thefuel elements are made of natural uranium in a magnesium-aluminum alloy (Magnox) anda full core consists of 50 MTU. We have based most of our historical information regardingthis reactor and its power history on figure VI.2 in Nuclear Puzzle . The 5 MW e reactor beganoperation in January 1986, experienced a 70 day shutdown in 1989, and continued irradiationpast the first IAEA inspections in 1992 until the shutdown in April 1994. We are primarilyconcerned with two possible fueling histories: case I, or the no-diversion case, assumes thatthe same core was used in the 5 MW e reactor during the entire irradiation period from 1986to 1994; case II, the core exchange case, is based on the assumption that, during the 70 dayshutdown of 1989, North Korea replaced the irradiated core with a fresh one and continued59rradiation with a higher-than-declared power to reach the same burn-up as in case I bythe time of the 1992 inspection. The primary safeguards-relevant difference between thesetwo cases is that in the second case an entire spent fuel load containing about 8.8 kg ofweapons-grade plutonium, is unaccounted for.We simulate the 5 MW e reactor using the SCALE 6.1.1 software package developedat Oak Ridge National Laboratory. OrigenArp, a subset of SCALE, uses decay data fromENDF/B-VII and neutron information from the JEFF/A-3.0 databases to computeburn-up and isotopic composition. OrigenArp is a deterministic approach which approxi-mates the structure of a reactor core into a zero-dimensional object by using appropriatelyweighted cross section libraries. Some libraries, including the Magnox type reactor, arepredesigned and supplied with the SCALE software.OrigenArp begins by computing depletion equations for each individual isotope in a givenproblem. The depletion or Bateman equation is given by dN i dt = m (cid:88) j =1 l ij λ j N j + ¯Φ m (cid:88) k =1 f ik σ k N k − ( λ i + ¯Φ σ i ) N i ( i = 1 , ..., m ) (18)This accounts for processes that produce nuclide N i in the first two terms and for processesthat destroy nuclide N i in the last, negative term. The first term represents the decays ofnuclide j to i given by the decay constant, λ j , the atom density of nuclide j , N j and thebranching fraction, l ij , for decays from nuclide j to i . The second term indicates neutroncaptures into nuclide i given by the space and energy-averaged neutron flux, ¯Φ, the fractionof absorption on nuclide k that produce nuclide i , f ik , and the spectrum-averaged neutronabsorption cross section of nuclide k , σ k . The last term is the collection of depletion modesconsisting of the decay of nuclide i via decay constant, λ i , and neutron absorption with aspectrum-averaged neutron absorption cross section of nuclide i , σ i . The indices are summed .113. “JEFF database.” able IX: Comparison of our SCALE-based results and the numbers presented in Plutonium 1996 for the plutonium content of Magnox fuel as a function of burn-up.SCALE
Plutonium 1996
SCALE
Plutonium 1996
Burn-up [MWd/t] % , , Pu % , , Pu kg of Pu kg of Pu100 0.99 0.75 0.10 0.1200 1.9 1.5 0.20 0.19300 2.9 2.3 0.29 0.28400 3.8 3.1 0.38 0.36500 4.7 3.7 0.47 0.45600 5.5 4.4 0.56 0.535700 6.4 5.1 0.64 0.62800 7.2 5.7 0.72 0.7900 8.0 6.3 0.79 0.781000 8.8 6.9 0.87 0.861100 9.5 ∼ ∼ ∼ ∼ over all branches including nuclide i . SCALE solves this differential equation via a matrixexponential method. Short-lived isotopes are removed to prevent loss of numerical accuracyand calculated using Bateman chains. With the input of initial nuclide concentrations, thepower history, and the reactor configuration, OrigenArp can provide time-dependent fissionrates, radioactivity, and isotopic abundances during and after irradiation.For our calculations, we have used the Magnox library provided with SCALE. Our firstcheck of SCALE is to investigate the production of plutonium in comparison to earlierresults. We irradiate the Magnox fuel corresponding to 1 MTU of natural uranium for 1000days at a constant power level to reach a given final burn-up. For example, we can irradiatethe core at 0.7 MW for 1000 days to produce a final burn-up of 700 MWd/t. We then extractthe total amount of plutonium produced and the percentage of
Pu, plutonium-241, and
Pu. table IX summarizes the results and compares them with table A.2 in
Plutonium1996 . We can see that our calculation and the results of
Plutonium 1996 consistentlymatch in the total mass of plutonium produced for various burn-ups. SCALE predicts about
Plutonium and highly enriched uranium 1996:world inventories, capabilities and policies (Oxford University Press, 1997).115. The numbers in
Plutonium 1996 are originally taken from ()
Plutonium 1996 .We also perform a comparison of the fissile abundances for the four main fissile isotopeswith the numbers quoted in table VIII.5 in
Nuclear Puzzle . We use the OrigenArp sequencewith the Magnox reactor library and the power history inferred from figure VI.2 of
NuclearPuzzle with an initial fuel amount of 1 MTU. The x-axis of the plot shows the number of daysthat have passed after the 5 MW e reactor began irradiation on January 1, 1986. Notabletimes are included, such as the 70 day shutdown in 1989 (beginning on t = 1156 d), the firstIAEA inspection in 1992 ( t = 2337 d), and the shutdown on April 1, 1994 ( t = 3012 d). Inthe leftmost panel of figure 11 we have plotted the results for case I and case II, as well asthe data points found in table VIII.5 in Nuclear Puzzle , normalized to a 50 MTU core andwe find excellent agreement. We note that immediately following the 70 day shutdown in1989, the fissile abundances vary greatly between cases. The difference quickly disappearsas the fresh fuel load is burned at a higher power so that by the first IAEA inspection in1992 the fissile abundance differences have vanished. The results in terms of mass inventoryand fission rates are shown in figure 11.
B CANDU and LEU reactors
The “H O, LEU” reactor and the “D O, NU” CANDU reactor are both calculated in thesame fashion as the 5 MW e reactor. The LEU reactor calculation is done for a typicalpressurized light water reactor. Specifically, we have taken a power history from one suchreactor, namely Ling Ao I, located in the Daya Bay complex in China. Ling Ao I is aFramatome M310 reactor, which uses a 17x17 AFA 3G fuel assembly. SCALE does nothave this specific library, but does contain the very similar Westinghouse 17x17 array, whichwe have used. Details of the Ling Ao I reactor history and fuel composition are takenfrom the yearly power histories published via IAEA Operation Experience in Member States62ocuments. To summarize, Ling Ao I has a total fuel load of 72.4 MTU enriched to 3.7%.The yearly power histories are converted into OrigenArp input files. The Ling Ao I reactorruns on a 12 month refueling cycle, meaning that it must shut down about every 12 monthsto refuel. Typically, one third of the fuel is replaced with a fresh third, and the fuel rodsare shuffled within the core to reach a flatter burn-up distribution. To simulate this, weproduce three SCALE computations: one third of a core that has been irradiated once,another third that has been irradiated twice, and the last third that has been irradiatedthree times. These three output files are then summed resulting in the final full core. Eachthird has been irradiated for approximately 335 days per cycle at an average power of about965 MW th , resulting in a total power of 2.9 GW th . We acquire the fission rates and the fissileabundances for the four main fissiles from this final full core sum. The “D O, NU” is alsocalculated via SCALE. We use the CANDU 37-element cross section library in SCALE. The37-element, as opposed to the 28-element, design was chosen as it is a more common designfor newer CANDU reactors. This simulation was performed with a three year irradiation timeat an average power of 40 MW th with a 8.6 MTU natural uranium fuel load. The CANDUreactor is run continuously with no refueling periods. From the SCALE output we obtainthe fission rates and the fissile abundances for the four main fissiles. The specifications forthis calculation are intended to mirror the specifications of the Arak reactor in Iran. C The IRT reactor
The IRT is a light-water pool-style research reactor and was supplied to the DPRK bythe Soviet Union in the 1960s. First criticality occurred in the IRT reactor on August 15,1965. The IRT contains 56 core grid compartments during the time of interest here, i.e.after 1986.
The exact configuration of the IRT is unknown, but we can base possible .117. Albright and Walrond,
Update on the Arak Reactor .118. Albright and O’Neill,
Nuclear Puzzle . located in Bulgaria. Thesecompartments can contain a driver or target element. Drivers are the primary fission sourcein this research reactor and are made of highly enriched uranium. The targets, which areprimarily composed of fertile isotopes, such as natural uranium, will experience a smallnumber of fissions. In 1974, the IRT was upgraded in power from 2 MW th to 4 MW th andlater in 1986 it was upgraded from 4 MW th to 8 MW th . The driver element enrichmentalso increased during this time from 10% in 1967 to 80% by 1986. Exact inventories areunavailable, but estimates indicate the DPRK may have had access to at least 92 of these80% enriched driver elements.
From 1986 on-wards, 30 driver elements were typicallyloaded and the IRT could run for about 250 days out of the year.SCALE computations for the IRT are more difficult as there is no available cross sectionlibrary provided with SCALE. Therefore, we use the Triton and NEWT modules to producecustom cross section libraries for the IRT calculations. NEWT generates the neutron trans-port calculation for a user-defined core configuration, which can then be used by Triton overa sample burn-up history to produce decay and cross section libraries. The input informationwe provide consists of detailed isotopic compositions of the driver and target elements as wellas physical parameters of these elements and a core configuration. Information concerningthe driver and target elements is taken from table VIII.6 in
Nuclear Puzzle . To summarize,we are using 80% enriched U-Al alloy drivers with an aluminum cladding in a light-watermoderator surrounded by a reflector. The target elements are natural uranium metal in analuminum cladding. We utilize SCALE and the IRT power history provided in table VIII.7in
Nuclear Puzzle along with the initial loading of 6 kg for the drivers and 633 kg for thetargets.The actual core configuration is not known and detailed designs are unavailable so wewill first determine the impact of the unknown core configuration on the fission rates, as
Directory of Nuclear Reactors , vol. V - Research, Test and Experimental Reactors (Vienna,Austria: International Atomic Energy Agency, 1964).120. Albright and O’Neill,
Nuclear Puzzle .121. Ibid. Irradiation time (cid:64) days (cid:68) F i ss i on s (cid:64) s e c (cid:45) (cid:68) (cid:64) days (cid:68) F i ss il eabundan c e (cid:64) k g (cid:68) Figure 12: IRT fission rates of a lone driver and target element separately and their sum (leftpanel). The abundances of fissile isotopes in the IRT as calculated by the lone sum method (righthand panel). these rates determine the neutrino spectrum. We tested several methods of computing thefission rates of the IRT research reactor. The first group of methods considered a calculationof a full IRT core with both drivers and targets present. Two core configurations were testedand are illustrated in figure 14.The second group of methods consisted of calculating the drivers and targets separatelyand then summing the fission rates and fission yields for a full core. For this group weconsidered three sub-cases. We calculated one lone driver and one lone target separately,multiple drivers and multiple targets separately in a first core configuration and multipledrivers and multiple targets separately in a second, different, core configuration. The specificdimensions of each element and the two different core configurations can be found in figure 13and figure 14. For the multiple drivers and targets separately we simply removed either thetarget or driver elements in figure 14.The fission rates across these several methods were found to be nearly identical. The fullcore of both targets and drivers produced slightly higher plutonium-239 and uranium-23865ssion rates. This is a result of the additional fission neutrons produced from the interplaybetween drivers and targets, which combines with the total neutron flux. This increase is avery small effect, especially when compared to the uranium-235 fission rate, which dominatesthe IRT. From this, we conclude that a specific core configuration has only a minor impacton the fission rates and we use the lone element and lone driver calculations. The primaryreasoning behind this is to avoid any unverifiable core configuration bias. The fission ratesfor the lone element method are given in figure 12. Blue curves indicate fission rates for thedriver, red curves indicate fission rates for the targets and purple curves are the sum of these.We have only illustrated the three main fissiles that contribute to the overall fission rates.The drivers comprise the majority of uranium-235 fissions and, thus, the majority of fissionsover all fissiles. The addition of targets increases the plutonium-239 and uranium-238 fissionrates, but these are still an order of magnitude lower than the uranium-235 fission rates.Again, we see the burn-up effect by the decreasing fission rate of uranium-235 along with anincrease in the plutonium-239 fission rate.We also wish to track the abundances of the four main fissile isotopes as a function of timeover the given 250 day irradiation cycle of the IRT. The abundances have been calculatedusing SCALE in the same fashion as described before. We attempt three different methods,a lone driver and lone target core individually calculated and then summed, a full core withdrivers and a full core with targets individually calculated and then summed, and a fullcore containing both drivers and targets. Again, the differences in these three methods wereextremely minor resulting in the similar conclusion that the core configuration is negligiblewhen considering the fissile abundances for the IRT. In the right panel of figure 12 we comparethe fissile abundances as produced via the lone target and element separately calculated andthen summed to a full core with table VIII.7 in
Nuclear Puzzle . We find that there is anexcellent match between the two results. The largest deviation occurs in plutonium-241 asthere is little of this isotope being produced.To summarize, we conclude that all methods we tested for computing the fission rates,66sotopic abundances, and neutrino fluxes for the IRT yield very similar results. The maineffect in the IRT is the addition of breeding targets and we reproduce previous results in theliterature. abcdefg ModeratorAl CladdingU (cid:45)
Al FuelU (cid:45) Al (cid:72) (cid:76) Al (cid:72) (cid:76) a (cid:61) (cid:61) (cid:61) (cid:61) (cid:61) (cid:61) (cid:61) ab c ModeratorAl CladdingU TargetAl (cid:72) (cid:76) a (cid:61) (cid:61) (cid:61) Figure 13: The left hand panel shows the IRT driver element material diagram, whereas theright hand panel shows IRT target element material diagram. a b c
ModeratorAl CladdingU (cid:45)
Al FuelU TargetReflector a (cid:61) (cid:61) (cid:61) a b c ModeratorAl CladdingU (cid:45)