North Korea's Plutonium Infrastructure
What nuclear fuel cycle modelling and open sources tell us about North Korea’s nuclear ambitions
The Verification Research, Training, and Information Centre (VERTIC), the James Martin Center for Nonproliferation Studies (CNS), and the Royal United Services Institute (RUSI) have combined fuel cycle modelling and open source information to explore different scenarios for the DPRK’s current inventory of weapon-usable material, and how that inventory might change. This story explores how the DPRK can generate plutonium in its reactors, and how it can separate out that plutonium from spent reactor fuel to make it useable in its weapons programme. It does this by combining open sources of information regarding reactor characteristics and operational histories with fuel cycle modelling to create estimates of the DPRK's plutonium inventory.
Plutonium production in the DPRK
Plutonium is generated by bombarding uranium with neutrons. When atoms of uranium-238 capture a neutron, they become unstable and decay via neptunium into plutonium. The DPRK has three potential neutron sources that they could exploit for this purpose.
The 5MWe Reactor
The 5 MWe reactor is a Magnox-type reactor located at Yongbyon and has been operating intermittently since 1986. The reactor is powered by natural (unenriched) uranium, and can accommodate 50 tonnes of metallic uranium fuel. The reactor is well-suited to generate weapons-usable plutonium in its fuel, particularly in comparison to other available neutron sources.
The IRT-DPRK Research Reactor
The Soviet Union played an important role in the DPRK's early nuclear development. It trained a number of DPRK scientists, and also provided the DPRK in the 1960s with an 'IRT'-type research reactor. The DPRK has accumulated a stockpile of spent IRT fuel, and is known to have reprocessed small quantities of that fuel in the 1970s. The plutonium content in this fuel is likely to be much lower than spent fuel from the 5MWe reactor, and will have an isotopic composition that makes it less appealing for weapons use. For the purposes of this model, it is assumed that spent fuel from the IRT reactor is not reprocessed and none of the plutonium generated in the fuel is available for the DPRK's plutonium stockpile. The neutron flux in the reactor may be used to bombard separate targets of uranium-238 in the experimental channels, but the quantities generated (relative to other sources) are considered insignificant and are not modelled.
The Experimental Light Water Reactor (ELWR)
The DPRK has claimed that the ELWR will be used primarily for electricity production. Light water reactors typically consume their fuel to a far greater extent than the plutonium-producing 5MWe reactor - approximately 22,000 Thermal Megawatt (MWth) days per ton, compared to 600-700 MWth days per ton. If this were the case for the ELWR, the plutonium it generates would have an isotopic composition that makes it less attractive (but not unavailable) for weapons use. As the 5MWe plutonium-producing reactor ages, the DPRK might look to the ELWR as an alternative source of plutonium for its weapon programme – even if it had to adjust its weapon programme to accommodate less attractive plutonium, or interrupt electricity production to refuel the reactor more frequently. In this case, the DPRK would also have to modify its reprocessing infrastructure to account for the different chemical composition and physical form of the ELWR fuel. There is no compelling evidence to suggest that the ELWR is operational, and if it becomes operational any separation of plutonium from spent fuel may be some time away. While fuelling the ELWR may place a near-term demand on the DPRK’s uranium stockpile and enrichment capacity (see uranium story here ), it is not an immediate source of weapons-usable plutonium.
The 5MWe Reactor
The 5 MWe reactor is a Magnox-type reactor located at Yongbyon and has been operating intermittently since 1986 [1, 2]. The reactor was subject to IAEA safeguards during some of its early operational phases, and its operation was controlled and restricted under both the Agreed Framework and the Six-Party Talks. As such, detailed information is available to the IAEA and some states regarding its initial design, its operational parameters, and the amount of nuclear material involved in its operation. Some of this information has been made publicly available through the IAEA, through transparency visits hosted by the DPRK, and through the DPRK's own announcements. Since the collapse of the Six-Party Talks, information relating to the reactor and its design has become more scarce. Satellite imagery has become a valuable source of information, indicating its likely operational windows and changes in coolant approaches (see Figure 1). Nevertheless, there is still uncertainty over the precise operating windows of this reactor, as demonstrated by Figure 2, which sets out individual claims from a variety of open sources and assigns a numerical value to the likelihood of reactor operation made in each claim. A sliding window below the chart can be used to focus on a date range.
Figure 1: The cooling water discharge is used as a signature alongside vehicle activity and thermal imagery (when available) to assess whether the 5 MWe Magnox-type reactor is operating.
Figure 2: Open source claims regarding 5MWe reactor operations. The grey bars represent the operating windows that were input to the model.
Core Loading
The reactor is powered by natural (unenriched) uranium, and can accommodate 50 tonnes of metallic uranium fuel. The IAEA describes the reactor as having six operational cycles, with the first cycle consisting of operations between 1986-1994 and the sixth cycle having commenced in July 2021 [1]. In the first operational cycle there is still uncertainty about when loading and unloading occurred, how much fuel was unloaded, and exactly how much fuel was reprocessed [2]. However, DPRK officials have since explained that while individual fuel elements can be charged or discharged while the reactor is operating, its core is typically fully unloaded and reloaded at the end of each cycle [3]. This model assumes that the DPRK requires 50 tonnes of natural uranium for each operational cycle. Magnox-type reactors have known issues such as warping of the graphite moderator, which are expected to degrade reactor performance over time.
Burn-up
DPRK officials have also explained that to preserve a desirable isotopic balance of plutonium, they restrict the irradiation of the fuel to below 3000 MWth days per ton of fuel [3]. According to those officials, fuel is typically discharged when the burnup reaches 600-700MWth-d per ton. With that burn-up limit in mind, they have argued that the core is refuelled every two to three years. Based on the observed operational history, and some declared information from visits and prior inspections, the cumulative estimated plutonium production for the reactor ranges from 25-48 kg by Hecker (in 2021) [4] and 31-35 kg up to 2016 by Albright [5]. A reactor core model estimates 46.7-76.6 kg of plutonium produced up to the end of the 2019 campaign [6].
Reprocessing
The radiochemical laboratory (RCL) at Yongbyon (Figure 3) is the only known reprocessing facility in the DPRK. It separates plutonium and depleted uranium from spent 5MWe reactor fuel via the 'PUREX' process - shearing the fuel rods, chemically stripping the Magnox cladding, and dissolving the nuclear material in its core in nitric acid and TriButyl Phosphate (TBP). Plutonium and uranium are then extracted from the dissolved fuel through a series of mixer-settlers, before the two elements are split through pulse columns. According to DPRK officials [3], plutonium nitrate solutions are converted directly into a plutonium metal alloy for use in the weapons programme. The depleted uranium nitrate liquors are kept in that form as a process waste, in a form that is not directly usable without further processing. Any volatile fission products (such as Iridium-131) are discharged directly to the atmosphere during the initial dissolution of spent fuel, with any other undesirable fission products remaining in liquid wastes.
Figure 3: The Radiochemical Laboratory at Yongbyon
Operational History
The first North Korean reprocessing that we are aware of took place in 1975, when the DPRK separated a small amount of plutonium from spent IRT fuel at the nearby Isotope Production Laboratory [2]. Reprocessing on a larger scale was not possible until the reprocessing lines at the RCL were complete. The first process line was tested in 1990 and the second completed in 1994. The two processing lines have an estimated capacity of 110 tonnes of uranium (or 220-250 tonnes depending on the source) [7]. This is far more than the required capacity for the 5 MWe Magnox-type reactor of 50 tonnes per year. There is still uncertainty about the plutonium produced from reprocessing campaigns prior to the Agreed Framework [1] that could be resolved with techniques such as nuclear archaeology [8]. Since the Agreed Framework, the IAEA has noted five full reprocessing campaigns, in 2003, 2005, 2009, 2016 and 2021 and a shorter campaign in 2018 [1]. Satellite imagery has played a valuable role in identifying these campaigns, for example by identifying when a supporting coal-fired power plant is operational (Figure 4). Figure 5 below sets out individual claims from a variety of open sources regarding the operational history of the RCL, and assigns a numerical value to the likelihood of reprocessing operations made in each claim. For the purpose of this model, reprocessing campaigns have been extrapolated from these claims and overlayed in the chart. A sliding window below the chart can be used to focus on a date range.
Figure 4: Effluent from the smokestack of the thermal plant which powers the Radiochemical Laboratory has been used as a signature to determine whether a reprocessing campaign is taking place, as has the accumulation of coal ash in an adjacent holding pond and its removal.
Figure 5: Open source claims regarding Radiochemical Laboratory operations. The grey bars represent the operating windows that were input to the model.
Operational Efficiency
The post-Agreed Framework reprocessing campaign in 2003 is especially notable for high corrosion of fuel from early reactor operations, due to it being stored in the Magnox pond from 1994-2002 which is expected to lead to significant material losses in the de-cladding stage for this fuel load [9]. This is the largest loss stage in reprocessing, with the subsequent stages of the PUREX process typically being very efficient. The overall losses from reprocessing, with most of the losses occurring at the de-cladding stage and some liquid stage losses of plutonium and reprocessed uranium to waste streams are modelled at 15%. Subsequent fuel loads were reprocessed sooner after discharge from the reactor reducing these losses but are modelled currently with the same loss rate [10]. The overall losses from reprocessing, with most of the losses occurring at the de-cladding stage and some liquid stage losses of plutonium and reprocessed uranium to waste streams are modelled at 15%. Subsequent fuel loads were reprocessed sooner after discharge from the reactor reducing these losses but are modelled currently with the same loss rate.
Plutonium stockpiles
First-hand claims by DPRK officials regarding the operational characteristics of the 5MWe reactor and visible satellite signatures of its operation provide a relatively sure footing from which to model the likely generation of plutonium in that reactor. While the reactor is still operational and the DPRK is available to fuel it, it is reasonable to assume that it will be operated in a way to maximise the production of weapons-usable plutonium as claimed by DPRK officials. It is more challenging to identify reprocessing campaigns and the efficiencies of those campaigns via open sources. However, given the visible signatures of operation and the known capacity of the RCL, the majority of plutonium generated in spent 5MWe fuel can be extracted and added to the DPRK's plutonium stockpile (barring some operational inefficiency losses). The possible growth of the DPRK's plutonium stockpile through individual reprocessing campaigns is illustrated in Figure 6 below.
Figure 6: Modelled estimate of the DPRK's stockpile of separated plutonium
Footnotes
[1] IAEA. Application of Safeguards in the Democratic People's Republic of Korea. GOV2022/40-GC(66)/16. Available here .
[2] DAVID ALBRIGHT & KEVIN O'NEILL. 2000. Solving the North Korean nuclear puzzle, Institute for science and international security Washington, DC.
[3] CHAIM BRAUN, SIEGFRIED HECKER, CHRIS LAWRENCE, PANOS PAPADIAMANTIS, North Korean Nuclear Facilities After the Agreed Framework, Center for International Security and Cooperation, Stanford University, 2016.
[4] SIEGFRIED HECKER. Estimating North Korea’s Nuclear Stockpiles: An Interview With Siegfried Hecker. 38 North. Available here . [Last Updated 2021]
[5] DAVID ALBRIGHT. North Korea’s Nuclear Capabilities: A Fresh Look. Institute for Science and International Security. Available here . [Last Updated 2017]
[6] GEON HEE PARK & SER GI HONG. An estimation of weapon-grade plutonium production from 5 MWe YongByon reactor through MCNP6 core depletion analysis. Progress in Nuclear Energy, 130 (2020) 103533
[7] WHANG JOOHO & GEORGE T BALDWIN. Dismantlement and radioactive waste management of DPRK nuclear facilities. SNL-A (Sandia National Laboratories, Albuquerque, NM). 2005.
[8] JULIEN DE TROULLIOUD DE LANVERSIN & MORITZ KÜTT. Verifying North Korea’s Plutonium Production with Nuclear Archaeology. Science & Global Security, 29 (2021) 145-166
[9] A private consultation from an expert familiar with the PUREX process estimated very high losses at the decladding stage.
[10] Based on private remarks at a 2019 workshop.
Credits
This work is part of an ongoing joint project between the Verification Research, Training and Information Centre (VERTIC), the James Martin Centre for Nonproliferation Studies (CNS) and the Royal United Services Institute (RUSI), and is funded by Global Affairs Canada. We also thank the UK National Nuclear Laboratory (NNL) for use of Orion software.
This model uses Orion software, which has been developed by the National Nuclear Laboratory Ltd. The results of the model are driven by assumptions and data selected by the authors and therefore should only be viewed as an indication the DPRK fuel cycle.
