Education

Research Description

Dr. Hou's research aims to improve reactor designs issue through two main avenues. The first involves developing cutting-edge modeling and simulation (M&S) capabilities in a true multiphysics context. The second focus on integrating advanced artificial intelligence (AI) and machine learning (ML) techniques with design and operational experiences. M&S tools are designed with a deliberate balance between accuracy and efficiency, in order to satisfy the varying demands at different stages of technology development. A significant advancement is the validation of AI/ML models using extensive organic and high-fidelity simulation data. These models are crucial for guiding nuclear power system design, operation and maintenance (O&M), ultimately enhancing their robustness and boosting competitiveness.

Publications

Development and application of two-step uncertainty propagation and sensitivity analysis methodology for fast reactor safety analysis

Trivedi, I., Delipei, G., Hou, J., Grasso, G., & Ivanov, K. (2025), NUCLEAR ENGINEERING AND DESIGN, 433. https://doi.org/10.1016/j.nucengdes.2025.113882

A System Predictive Maintenance Framework for Advanced Reactors Using a Data-Driven Digital Twin

Rivas, A., Delipei, G. K., & Hou, J. (2024, August 3), NUCLEAR SCIENCE AND ENGINEERING, Vol. 8. https://doi.org/10.1080/00295639.2024.2372515

A component diagnostic and prognostic framework for pump bearings based on deep learning with data augmentation

Rivas, A., Delipei, G. K., Davis, I., Bhongale, S., Yang, J., & Hou, J. (2024), RELIABILITY ENGINEERING & SYSTEM SAFETY, 247. https://doi.org/10.1016/j.ress.2024.110121

A system diagnostic and prognostic framework based on deep learning for advanced reactors

Rivas, A., Delipei, G. K., Davis, I., Bhongale, S., & Hou, J. (2024), PROGRESS IN NUCLEAR ENERGY, 170. https://doi.org/10.1016/j.pnucene.2024.105114

Development and evaluation of parallel simulated annealing algorithm for reactor core optimization problems

Mikouchi-Lopez, J., Delipei, G., & Hou, J. (2024), Nuclear Science and Technology Open Research. https://doi.org/10.12688/nuclscitechnolopenres.17464.2

Modeling a generic Pebble Bed High-Temperature Gas-Cooled Reactor to perform load-following using Simulink

Rivas, A., Delipei, G. K., & Hou, J. (2024), NUCLEAR ENGINEERING AND DESIGN, 428. https://doi.org/10.1016/j.nucengdes.2024.113506

Operation Optimization Framework for Advanced Reactors Using a Data-Driven Digital Twin

Rivas, A., Delipei, G. K., & Hou, J. (2025), JOURNAL OF NUCLEAR ENGINEERING AND RADIATION SCIENCE, 11(2). https://doi.org/10.1115/1.4066777

Prototyping of a Machine Learning-Based Burnup Measurement Capability for Pebble Bed Reactor Fuel

Rollins, N., Allan, I., & Hou, J. (2025). Prototyping of a Machine Learning–Based Burnup Measurement Capability for Pebble Bed Reactor Fuel. Nuclear Science and Engineering. https://doi.org/10.1080/00295639.2024.2328937,

Reinforcement Learning-Based Control Sequence Optimization for Advanced Reactors

Nguyen, K. H. N., Rivas, A., Delipei, G. K., & Hou, J. (2024), JOURNAL OF NUCLEAR ENGINEERING, 5(3), 209–225. https://doi.org/10.3390/jne5030015

Source term analysis of FeCrAl accident tolerant fuel using MELCOR

Baker, U., Choi, Y.-J., Rollins, N., Nguyen, K., Jung, W., Whitmeyer, A., … Lindley, B. (2024), ANNALS OF NUCLEAR ENERGY, 202. https://doi.org/10.1016/j.anucene.2024.110482

View all publications via NC State Libraries

View publications on Google Scholar

Grants

Development of LWR Fuel Reload Optimization Framework

Battelle Energy Alliance, LLC(10/28/21 - 1/31/25)

×

Development of LWR Fuel Reload Optimization Framework

Sponsor: Battelle Energy Alliance, LLC

Start Date: 10/28/21

End Date: 1/31/25

Abstract

This proposed work intends to support the ongoing development within the framework with a target effort to identify and select the optimal reactor core M&S tools to be integrated and to further develop various key capabilities of the framework. More specifically, the objectives of the study are twofold: 1) Test and assess features and modeling capabilities of various multi-physics reactor core simulators through well controlled benchmark calculations. 2) Develop and enhance the automation of core design optimization in RAVEN within the Plant Fuel Reload Optimization Framework.

Close

Machine Learning Based Lattice and Core Optimization Methodology for Multi-Cycle Operations ��� CNP Core Project 6

Consortium for Nuclear Power (CNP)- Dept of Nuclear Engineering(7/01/22 - 6/30/24)

×

Machine Learning Based Lattice and Core Optimization Methodology for Multi-Cycle Operations ��� CNP Core Project 6

Sponsor: Consortium for Nuclear Power (CNP)- Dept of Nuclear Engineering

Start Date: 7/01/22

End Date: 6/30/24

Abstract

One of the key challenges of the nuclear industry for operating the current nuclear power plants (NPPs) is to reduce the cost, often evaluated by the levelized cost of energy (LCOE), while maintaining safety and efficiency of nuclear reactors. The proposed project aims to develop and apply ML/AI RL techniques for the combined in-core and out-of-core optimization of PWR using the NCSU Modular Optimization Framework (MOF) . The Python 3 based MOF currently has two solvers, GA and SA, and can solver for assembly optimization, single cycle, and multi-cycle core optimization using the code interface with CASMO4 (lattice calculations) and SIMULATE3 (core simulation). CASMO4/SIMULATE3 is a computationally very efficient code suite that will allow the exploration of various optimization goals and constraints. The RL solver will be developed within MOF by leveraging existing capabilities to reduce the lead time.

Close

Performance Evaluation of Accident Tolerant Fuel in Optimized LWR Designs

US Dept. of Energy (DOE)(2/17/22 - 9/30/23)

×

Performance Evaluation of Accident Tolerant Fuel in Optimized LWR Designs

Sponsor: US Dept. of Energy (DOE)

Start Date: 2/17/22

End Date: 9/30/23

Abstract

This proposed work aims to develop an optimized core design for a generalized PWR with a two-year cycle length using ATF and to evaluate the core performance using high-fidelity multiphysics simulation code VERA-CS. Increased fuel enrichment and extended fuel burnup will be implemented to achieve the desired cycle length. Literature review of the ATF will be conducted first with the focus placed on the Iron-Chromium-Aluminum (FeCrAl) cladding. Metaheuristic optimization methods including genetic algorithm (GA) and simulated annealing (SA) available in the Modular Optimization Framework (MOF) will be leveraged to perform the optimization. Results from VERA-CS will be converted to be used in the MACCS (MELCORE Accident Consequence Code System) for the analysis of offsite consequences of a hypothetical release of radioactive material to the environment.

Close

Xe-100 Burn Up Measurement System, CNP Core Project #5

Consortium for Nuclear Power (CNP)- Dept of Nuclear Engineering(7/01/21 - 6/30/23)

×

Xe-100 Burn Up Measurement System, CNP Core Project #5

Sponsor: Consortium for Nuclear Power (CNP)- Dept of Nuclear Engineering

Start Date: 7/01/21

End Date: 6/30/23

Abstract

X Energy, LLC (X-energy) is transforming the nuclear energy marketplace through the development of the Xe-100, a 200 MWth modular pebble-bed HTGR to produce electricity and/or process steam. The Xe-100 Reactor uses pebble Type fuel elements (6 cm in diameter) which contains around 19,000 Triso-coated particles each. The Xe-100 Reactor utilizes an online refueling fuel handling system to achieve an availability factor of 95%. On average the pebble fuel will pass the Xe-100 reactor an average of 6 times to achieve the design burn-up of 165,000 MWd/tHM. During the start-up procedure of the Xe-100 reactor different enrichment pebbles will be introduced to the Xe-100 reactor until the equilibrium core is achieved with an enrichment of 15.5%. It is important for the Xe-100 Burn-Up-Measurement-System (BUMS) which forms part of the Xe-100 fuel handling system to distinguish between both pebble fuel burn-up and enrichment level. For the CNP Project, X-Energy proposes to develop an advanced methodology to continuously measure and analyze fuel pebbles discharge from the Xe-100 HTGR core to achieve the objective of burnup and enrichment measurement based on the gamma and neutron spectrometry. It is proposed to adopt a two-step approach with the first step focusing on distinguishing the influence of burnup and enrichment level of the pebble fuel, and the second step aiming at the fuel burnup estimation. Machine learning models will be developed, first using synthetic data and later using experimental results, to achieve accurate and real-time analysis. The method will be demonstrated on the detection system being engineered at X-energy.

Close

Machine Learning Model Development in 2022 (ARPA-E WP 2.3.3)

US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)(1/03/22 - 12/31/22)

×

Machine Learning Model Development in 2022 (ARPA-E WP 2.3.3)

Sponsor: US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)

Start Date: 1/03/22

End Date: 12/31/22

Abstract

This project is aims for the development and application of machine learning (ML) models to the nuclear power plant diagnosis and prognosis, and predictive maintenance, to support the reduction of operation and maintenance (O&M) cost of advanced reactor concepts. The first phase is focused on the development of a ML model for pump degradation prediction and failure mode diagnosis, more specifically aiming to make remaining useful life (RUL) predictions and corresponding failure mode of pumps. A synthetic database for the pump operation (and degradation) over time will be generated using the physics model mentioned above. Next, a ML model will be trained, tested, and validated for the pump degradation prediction and failure mode diagnosis using the sensor readings synthesized based on the pump degradation database. In the second phase, we will integrate the physics-based pump degradation model developed in Phase 1 with the Xe-100 simulator and develop a ML degradation model that is capable of making predictions using system-wise sensor data.

Close

Machine Learning Methods for Xe-100 Plant Monitoring and Operations (ARPA-E WP 2.3.2)

US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)(9/01/21 - 12/31/22)

×

Machine Learning Methods for Xe-100 Plant Monitoring and Operations (ARPA-E WP 2.3.2)

Sponsor: US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)

Start Date: 9/01/21

End Date: 12/31/22

Abstract

In the proposed project, we seek to develop advanced plant monitoring and operating technologies for the Xe-100 by developing, implementing, and validating the machine learning (ML) based algorithms. It will be carried out in two sub-projects, each concentrating on one focus area. First, we will develop an advanced methodology to continuously measure and analyze fuel pebbles discharge from the Xe-100 HTGR core to achieve the objective of burnup and enrichment measurement based on the gamma and neutron spectrometry. It is proposed to adopt a two-step approach with the first step focusing on distinguishing the influence of burnup and enrichment level of the pebble fuel, and the second step aiming at the fuel burnup estimation. ML models will be developed, first using synthetic data and later using experimental results, to achieve accurate and real-time analysis. The method will eventually be demonstrated on the detection system being engineered at X-energy.

Close

Machine Learning for ARPA-E GEMINA Project (ARPA E WP 2.3.1)

US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)(1/29/21 - 12/31/22)

×

Machine Learning for ARPA-E GEMINA Project (ARPA E WP 2.3.1)

Sponsor: US Dept. of Energy (DOE) - Advanced Research Projects Agency - Energy (ARPA-E)

Start Date: 1/29/21

End Date: 12/31/22

Abstract

"X-energy is proposing a project to significantly reduce the fixed operator and maintenance (O&M) cost of an advanced reactor to a target of $2/MWh. The project will follow two parallel paths to achieve its cost saving goals. The first path will access the plant using techniques such as human factors engineering, risk and hazard analysis, and maintenance evaluations to identify areas for optimization and then to develop new and innovative ways to leverage techniques such as automation, robotics, remote and centralized maintenance, and monitoring to optimized and reduce staff. A second path involves the development of virtual models that can be used to evaluate and validate the solutions proposed above. Two virtual models are an Immersive Environment Toolset and a Digital Twin. The Immersive Environment Toolset is a multidiscipline 3D model which, when combined with virtual reality, will be used to test maintenance concepts and techniques and optimize maintenance and security staff. The Digital Twin synthesizes information from the as operating plant, past operating history and planned future evolutions as well as assimilating data from other plants across the fleet to provide a wholistic understanding of the plant, health and remaining lifetime of critical plant components and identify predictive maintenance needs of physical assets with the overall goals of reducing staff requirements by optimizing O&M. The Xe-100 reactor features unsurpassed inherent safety characteristics which make it ideal to showcase the abilities of the Digital Twin. X-energy������������������s project will demonstrate an estimated 86% reduction in the Xe-100 plant fixed O&M Costs.

Close

Demonstration Of Utilization Of High-fidelity Neams Tools To Inform The Improved Use Of Conventional Tools Within The Neams Workbench On The 18-15104

US Dept. of Energy (DOE)(10/01/18 - 9/30/22)

×

Demonstration Of Utilization Of High-fidelity Neams Tools To Inform The Improved Use Of Conventional Tools Within The Neams Workbench On The 18-15104

Sponsor: US Dept. of Energy (DOE)

Start Date: 10/01/18

End Date: 9/30/22

Abstract

The primary goal of the proposed project is to demonstrate the utilization of high-fidelity Nuclear Energy Advanced Modeling and Simulation (NEAMS) tools (PROTEUS, Nek5000, and BISON)) to inform the improved use of conventional tools (DIF-3D, CTF, and CTFFuel) within the NEAMS Workbench on the NEA/OECD C5G7-TD benchmark. The project will highlight/illustrate some of the main objectives of the NEAMS workbench: to enable end users to use high-fidelity NEAMS tools to inform the improved use of lower-order conventional tools within the Workbench as well as to demonstrate the values of NEAMS tools as applied to collaborative benchmarks from a common input/template. Once demonstrated the proposed approach could also be extended and applied to different geometries and reactor types for structured and unstructured grids. The proposed project is envisioned as a partnership with the developers of the high-fidelity NEAMS tools and the NEAMS workbench and an industrial user of the tools. In order to make the project results relevant to industrial applications the focus will be on obtaining an optimal combination of accuracy and efficiency, which satisfies the current industry needs and performance requirements

Close

Development of Simulation Capabilities and Validation Basis to Support Evaluation of Advanced Nuclear Fuel Concepts

US Dept. of Energy (DOE)(10/01/20 - 3/11/22)

×

Development of Simulation Capabilities and Validation Basis to Support Evaluation of Advanced Nuclear Fuel Concepts

Sponsor: US Dept. of Energy (DOE)

Start Date: 10/01/20

End Date: 3/11/22

Abstract

The proposed project aims to support the ongoing development of advanced nuclear fuel design concepts, including the Accident Tolerant Fuel (ATF), HBU (High Burnup fuels), and High-Assay Low Enriched Uranium (HALEU), by extending selected computation tools, data, validation basis, and uncertainty analysis in the SCALE code system. These new fuel concepts have been proposed by the Department of Energy (DOE) and nuclear industry aiming to improve the safety and economics of nuclear power. The work intends to: 1) improve the energy deposition model and data (nominal value and its uncertainties) to be used in the lattice physics calculation; 2) extend the stochastic sampling capabilities to perform the uncertainty analysis of heat generation distribution in space and its variations in time; 3) develop a sampling-based similarity (or representation) evaluation method to allow for the validation and uncertainty quantification using existing validation database. The deliverables include documented methodologies, augmented data file, updated computational modules in the SCALE package, as well as verification and validation results.

Close

Development of High-Fidelity Transient Model for Pulsed Plasma Reactor

National Aeronautics & Space Administration (NASA)(10/01/20 - 2/28/21)

×

Development of High-Fidelity Transient Model for Pulsed Plasma Reactor

Sponsor: National Aeronautics & Space Administration (NASA)

Start Date: 10/01/20

End Date: 2/28/21

Abstract

In the proposed work, the NCSU team, led by Dr. Jason Hou, will provide technical support to the ongoing project: ����������������Pulse Plasma Rocket: Shielded, Fast Transits for Humans to Mars���������������, which is funded by the NASA Innovative Advanced Concepts (NIAC) Program. The lead Principal Investigator (PI) of the project is Steven Howe from Howe Industries LLC. The overall objective is to develop a propulsion system that may produce 20,000 lbsf of thrust with an Isp of 5,000 s. This work aims to support the ongoing modeling and simulation (M&S) effort of the nuclear reactor portion of the rocket concept using high-fidelity neutronics tools in the MOOSE framework. More specifically, the project intends to develop and refine a time-dependent simulation model for the reactor using the radiation transport code Mammoth/Rattlesnake, which takes the kinetics parameters and cross sections generated by the Monte Carlo code Serpent. Transient simulation of the core performance will be conducted using the reactor physics model and core design parameters will be perturbed to determine the optimal design.

Close