Momentum informed muon scattering tomography for monitoring spent nuclear fuels in dry storage cask.

Development of an effective monitoring method for spent nuclear fuel (SNF) in a dry storage cask (DSC) is important to meet the increasing demand for dry storage investigations. The DSC investigation should provide information about the quantity of stored SNF, and quality assurance of materials should be possible without opening the cask. However, traditional nondestructive examination (NDE) methods such as x-rays are difficult to deploy for DSC investigation because a typical DSC is intentionally designed to shield against radiation. To address this challenge, cosmic ray muons (CRMs) are used as an alternative NDE radiation probe because they can easily penetrate an entire DSC system; however, a wide application of muons is often hindered due to the naturally low CRM flux (~104 muons/m2/min). This paper introduces a newly proposed imaging algorithm, momentum-informed muon scattering tomography (MMST), and presents how a limitation of the current muon scattering tomography technique has been addressed by measuring muon momentum. To demonstrate its functionality, a commercial DSC with 24 pressurized light water reactor fuel assemblies (FAs) and the MMST system were designed in GEANT4. Three noticeable improvements were observed for MMST system as a DSC investigation tool: (1) a signal stabilization, (2) an enhanced capability to differentiate various materials, and (3) statistically increased precision to identify and locate missing FAs. The results show that MMST improves the investigation accuracy from 79 to 98% when one FA is missing and 51% to 88% when one-half FA is missing. The advancement of the NDE technique using CRM for DSC verification is expected to resolve long-standing problems in increasing demand for DSC inspections and nuclear security.

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients.

Nuclear reactor safety and efficiency can be enhanced through the development of accurate and fast methods for prediction of reactor transient (RT) states. Physics informed neural networks (PINNs) leverage deep learning methods to provide an alternative approach to RT modeling. Applications of PINNs in monitoring of RTs for operator support requires near real-time model performance. However, as with all machine learning models, development of a PINN involves time-consuming model training. Here, we show that a transfer learning (TL-PINN) approach achieves significant performance gain, as measured by reduction of the number of iterations for model training. Using point kinetic equations (PKEs) model with six neutron precursor groups, constructed with experimental parameters of the Purdue University Reactor One (PUR-1) research reactor, we generated different RTs with experimentally relevant range of variables. The RTs were characterized using Hausdorff and Fréchet distance. We have demonstrated that pre-training TL-PINN on one RT results in up to two orders of magnitude acceleration in prediction of a different RT. The mean error for conventional PINN and TL-PINN models prediction of neutron densities is smaller than 1%. We have developed a correlation between TL-PINN performance acceleration and similarity measure of RTs, which can be used as a guide for application of TL-PINNs.

Fieldable muon spectrometer using multi-layer pressurized gas Cherenkov radiators and its applications.

Cosmic ray muons have been considered as a non-conventional radiation probe in various applications. To utilize cosmic ray muons in engineering applications, two important quantities, trajectory and momentum, must be known. The muon trajectories are easily reconstructed using two-fold detector arrays with a high spatial resolution. However, precise measurement of muon momentum is difficult to be achieved without deploying large and expensive spectrometers such as solenoid magnets. Here, we propose a new method to estimate muon momentum using multi-layer pressurized gas Cherenkov radiators. This is accurate, portable, compact (< 1m3), and easily coupled with existing muon detectors without the need of neither bulky magnetic nor time-of-flight spectrometers. The results show that not only our new muon spectrometer can measure muon momentum with a resolution of ± 0.5 GeV/c in a momentum range of 0.1-10.0 GeV/c, but also we can reconstruct cosmic muon spectrum with high accuracy (~ 90%).

Monte Carlo characterization of the cosmic ray muon flux in shallow subsurface geological repositories intended for disposal of radioactive materials.

Recent challenges in monitoring subsurface geological repositories intended for disposal of radioactive materials such as spent nuclear fuel call for new, innovative concepts that are facility independent, cost-effective, passive, and reliable. Once nuclear material is in place at these facilities, reverifying the inventory may no longer be feasible if continuity of knowledge is lost or unavailable to the inspectors. Using cosmic ray muons may present several potential advantages over conventional photon/neutron signatures, and their use in safeguards applications have only received attention in the past decade. However, there have been limited efforts to explore the integration of cosmic ray muons into repository safeguards and study potential gains, risks, and costs. This paper presents a Monte Carlo-based methodology to characterize the cosmic ray muon flux, including muon angular and energy differential distributions at depths representative of subsurface geological repositories. Since there have been limited measurements at these sites and a measurement made in one site is not always transferable to another site, the objective is to develop an efficient simulation method and useful parametrizations to provide a convenient tool for enabling muon simulations at any geological repository site. It is expected these results will provide a better understanding of how muons can be integrated into an existing geological repository safeguards framework.

Muon Tracing and Image Reconstruction Algorithms for Cosmic Ray Muon Computed Tomography.

Cosmic ray muon-computed tomography (µCT) is a new imaging modality with unique characteristics that could be particularly important for diverse applications including nuclear nonproliferation, spent nuclear fuel monitoring, cargo scanning, and volcano imaging. The strong scattering dependence of muons on atomic number Z in combination with high penetration range could offer a significant advantage over existing techniques when dense, shielded containers must be imaged. However, µCT reconstruction using conventional filtered back-projection is limited due to the overly simple assumptions that do not take into account the curved path caused by multiple Coulomb scattering prompting the need for more sophisticated approaches to be developed. In this paper, we argue that the use of improved muon tracing and scattering angle projection algorithms as well as an algebraic reconstruction technique should produce muon tomographic images with improved quality - or require fewer muons to produce the same image quality - compared to the case where conventional methods are used. We report on the development and assessment of three novel muon tracing methods and two scattering angle projection methods for µCT. Simulated dry storage casks with single and partial missing fuel assemblies were used as numerical examples to assess and compare the proposed methods. The reconstructed images showed an expected improvement in image quality when compared with conventional techniques, even without muon momentum information, which should lead to improved detection capability, even for partial defects.