Reconstructing North Korea's Plutonium Production History with Bayesian Inference-Based Reprocessing Waste Analysis

TL;DR

This study demonstrates that Bayesian Reprocessing Waste Analysis (BRAM) can reconstruct the operating history of North Korea's Yongbyon 5 MWe reactor from reprocessing-waste isotopic ratios, by inferring per-cycle burnup $BU$ and cooling time $CT$ using a Bayesian framework. The approach combines a Gaussian-process surrogate for fast isotopic predictions, a multivariate normal likelihood, and uniform priors, and employs Leave-One-Out cross-validation to discriminate between histories with different numbers of core discharges. Results show that, with appropriately broad priors, BRAM can identify true burnup values and distinguish between plausible reactor histories, while wrong priors reveal inconsistencies that could flag false declarations. The work also illustrates how BRAM can yield a weighted plutonium production estimate reflecting uncertainty over alternative histories, and it advocates integrating BRAM with other nuclear-archaeology methods to strengthen verification credibility, pending experimental validation.

Abstract

Although North Korea's nuclear program has been the subject of extensive scrutiny, estimates of its fissile material stockpiles remain fraught with uncertainty. In potential future disarmament agreements, inspectors may need to use nuclear archaeology methods to verify or gain confidence in a North Korean fissile material declaration. This study explores the potential utility of a Bayesian inference-based analysis of the isotopic composition of reprocessing waste to reconstruct the operating history of the 5 MWe reactor and estimate its plutonium production history. We simulate several scenarios that reflect different assumptions and varying levels of prior knowledge about the reactor. The results show that correct prior assumptions can be confirmed and incorrect prior information (or a false declaration) can be detected. Model comparison techniques can distinguish between scenarios with different numbers of core discharges, a capability that could provide important insights into the early stages of operation of the 5 MWe reactor. Using these techniques, a weighted plutonium estimate can be calculated, even in cases where the number of core discharges is not known with certainty.

Figures (6)

• Figure 1: Schematic of a fuel channel of the 5 MWe reactor as modelled with OpenMCromanoOpenMCStateoftheartMonte2015 and ONIXdetroulliouddelanversinONIXOpensourceDepletion2021.
• Figure 2: Inference results of the test case with realistic priors. The posterior samples are illustrated by the blue histogram, the vertical orange line indicates the "true" parameter value and the red line indicates the respective prior interval.
• Figure 3: Inference results of the test case with wide priors. The posterior samples are illustrated by the blue histogram, the vertical orange line indicates the "true" parameter value and the red line indicates the respective prior interval.
• Figure 4: Inference results of the test case with wrong priors. The posterior samples are illustrated by the blue histogram, the vertical orange line indicates the "true" parameter value and the red line indicates the respective prior interval.
• Figure 5: Scenario comparison with LOO-CV. The x-axis represents two scenarios with distinct sets of simulated evidence. Based on the LOO-CV, each model is assigned one weight per isotopic ratio. The y-axis shows the average weight assigned to each of the two inference models per scenario.