Text

Ril 2021-16, Proceedings of the Workshop Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization
	ML21348A020
Person / Time
Issue date:	09/16/2021
From:	Carlson J, Doug Eskins, Gascot R, Raj Iyengar, Ulmer C; NRC/RES/DE
To:	;
	Gascot-Lozada R
Shared Package
ML21348A019	List: ML21348A020;
References
	RIL 2021-16
	Download: ML21348A020 (342)
	v • d • e

RIL 2021-16 PROCEEDINGS OF THE WORKSHOP ENABLING TECHNOLOGIES FOR DIGITAL TWIN APPLICATIONS FOR ADVANCED REACTORS AND PLANT MODERNIZATION Virtual Workshop September 14-16,2021 Date Published:

Prepared by:

J. Carlson D. Eskins R. Gascot R. Iyengar C. Ulmer U.S. Nuclear Regulatory Commission Research Information Letter Office of Nuclear Regulatory Research

Disclaimer This report was prepared as an account of work sponsored by an agency of the U.S. Government. Neither the U.S. Government nor any agency thereof, nor any employee, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third partys use, or the results of such use, of any information, apparatus, product, or process disclosed in this publication, or represents that its use by such third party complies with applicable law.

This report does not contain or imply legally binding requirements. Nor does this report establish or modify any regulatory guidance or positions of the U.S. Nuclear Regulatory Commission. This report is not binding on the Commission

Page intentionally left blank EXECUTIVE

SUMMARY

The Office of Nuclear Regulatory Research (RES) at the U.S. Nuclear Regulatory Commission (NRC) has initiated a future focused research project to assess the regulatory viability of digital twins for nuclear power plants. The objectives of this project are to:

Understand the current state of the technology and potential applications for the nuclear industry,

Identify and evaluate technical issues that could benefit from regulatory guidance, and

Develop infrastructure to support regulatory decisions associated with digital twins.

As a follow-on to the workshop hosted in December 2020 (ML21083A132), RES sponsored the Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization 2021 Online Workshop. The workshop was hosted by Idaho National Laboratory (INL) in collaboration with Oak Ridge National Laboratory (ORNL), the Department of Energys (DOE) Advanced Research Projects Agency-Energy (ARPA-E), and the Electric Power Research Institute (EPRI) and was held September 14-16, 2021.

The 3-day workshop was composed of five technical and panel sessions with 29 presenters from a wide range of national and international organizations, including universities, national laboratories, government agencies, nuclear vendors, nuclear industry, advanced reactor developers, and digital twin developers. With 324 participants from across the globe, the workshop provided a forum for nuclear industry and digital twin stakeholders to discuss the application of digital twins and digital twin enabling technologies such as advanced sensors and instrumentation, data analytics, machine learning and artificial intelligence in the current light water reactor (LWR) fleet and advanced reactor designs. The workshop also included an overview of the next steps toward regulatory realization of digital twins in the nuclear industry.

The workshop had two main purposes: (1) to review and exchange information on the current applications of digital twin enabling technologies, and (2) to identify necessary steps toward regulatory realization of digital twins. The workshop sessions covered the following topics: industry applications to digital twins in nuclear, advanced sensors and instrumentations, use cases of digital twin enabling technologies in nuclear power plants, digital twin enabling technologies in advanced reactor applications, and steps toward regulatory realization of digital twins.

On the first day of the workshop, Tuesday, September 14, 2021, Mr. Ray Furstenau, Director of RES, opened the workshop with introductory remarks and moderated a panel session on industry applications of digital twins in nuclear and Dr. Hasan Charkas from EPRI moderated a technical session on advanced sensors and instrumentations. On the second day of the workshop, Wednesday, September 15, 2021, Dr. Gene Carpenter representing Advanced Research Projects Agency - Energy (ARPA-E) moderated a technical session on use cases of digital twins enabling technologies in nuclear power plants and Ms. Angela Buford, Office of Nuclear Reactor Regulation (NRR), NRC, moderated a technical session on digital twin enabling technologies in advanced reactor applications. On the third day of the workshop, Thursday, September 16, 2021, Mr. Eric Benner, NRR, moderated a panel session on steps toward regulatory realization of digital twins and delivered the workshop closing remarks.

v

The following are some major takeaways from the workshop:

Technical Challenges/Opportunities

The nuclear industry and national laboratories have demonstrated interest and are pursuing the use of digital twin technologies and have realized advanced capabilities in preventive maintenance optimization, work order data analysis, anomaly detection and diagnosis, and real-time radiation monitoring.

There is significant interest/effort in the development of advanced sensor technologies and applications including wireless communication, multi-modal sensing, condition-based monitoring and maintenance, and operation in harsh environments, especially those introduced by advanced reactor designs (e.g., extreme temperature, radiation, corrosivity).

Many advanced reactor developers are designing plants integrated with digital twins (DTs) throughout their lifecycle to facilitate improved decision making and greater operational flexibilities (e.g., potential dynamic operating envelope).

Challenges exist in the following areas: real-time reduced-order or surrogate models, data production and integration, virtual prototyping, autonomous control, and sensor requirements.

Regulatory Challenges/Opportunities

There are three main categories of potential DT use: 1) use by industry for inherent benefits (e.g., improved design, construction, operations and maintenance), 2) use by industry as a tool for regulatory compliance (e.g., licensing submittals, safety analysis),

and 3) use as an NRC regulatory tool (e.g., shared source of plant information, enabler of iterative design approvals and just-in-time regulation).

Industry and regulators need to develop agreed upon guidance and frameworks for acceptance of DT applications that is consistent, explicit, and enables the use of DTs as an additional avenue for meeting the intent of existing regulations.

One approach to building confidence in DT technology - an important aspect for acceptance and adoption of DTs - is pioneering DT applications with non-safety components or systems and demonstrating acceptable performance prior to safety-related applications.

DTs have the potential to enhance NRC inspection activities, including automated regulatory compliance testing and on-demand access to high-fidelity plant information.

All presentations slides from this workshop are available in the NRCs Agencywide Documents Access and Management System, under Accession No. ML21342A121.

vi

Page intentionally left blank TABLE OF CONTENTS EXECUTIVE

SUMMARY

............................................................................................................v 1 Day 1 Presentations .............................................................................................................1 1.1 Panel Session 1: Industry Applications of Digital Twins in Nuclear ..................................1 1.1.1 Innovation Journey to a Brighter Future .............................................................1 1.1.2 EPRIs Digital Twin Related Activities for Nuclear Applications ..........................2 1.1.3 Westinghouse Perspectives...............................................................................2 1.1.4 Exelon Nuclear Innovation Projects Overview....................................................2 1.1.5 ARPA-E GEMINA Portfolio and Digital Twins ....................................................2 1.1.6 Digital Twin Monitoring for Advanced Reactors and Plant Modernization........... 2 1.2 Session 2: Advanced Sensors and Instrumentations .......................................................3 1.2.1 Advanced Sensors and Instrumentation for Digital Twin Applications ................ 3 1.2.2 Non-intrusive Temperature and Pressure Wireless Sensor and Transceiver System for Extreme Environment Applications ...................3 1.2.3 Data Analytics and Remote Monitoring Integration ............................................4 1.2.4 Digital Twin Impact on I&C Systems Development for Xe-100 ...........................4 1.2.5 Online Monitoring (OLM) Implementation to Extend Transmitter Calibration Intervals in Nuclear Facilities ................................................4 2 Day 2 Presentations .............................................................................................................5

2.1 Session

Use Cases of Digital Twin Enabling Technologies in Nuclear Power Plant ...........................................................................................................................5 2.1.1 Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform ...............................................................................5 2.1.2 AI Driven Scalable Condition-based Predictive Maintenance Strategy ............... 6 2.1.3 Thermal Performance Management for Nuclear Power Plant with Digital Twins .....................................................................................................6 2.1.4 Digital Twin of a Real-time Radiation Monitoring Network ..................................6 2.1.5 Non-destructive Examination (NDE) 4.0 and ML for In-Service Inspections .............................................................................................6 2.2 Session 2: Digital Twin Enabling Technologies in Advanced Reactors Applications ................................................................................................................ 7 2.2.1 Xe-100 Digital Twin Technologies Overview ......................................................7 2.2.2 Humble AI for Reliable Machine Learning-Based Health Twin ...........................8 2.2.3 Enabling Technologies for Digital Twins Applications for the KP-FHR ............... 8 2.2.4 High-Fidelity Digital Twins for BWRX-300 Critical Systems................................8 2.2.5 Molten Salt Loop Development Acceleration with Disturbed Single Crystal Harsh Environment Optical Fiber-Sensors .................................9 2.2.6 Digital Twin to Production Reactors, The Simulation Continuum .......................9 3 Day 3 Presentations ...........................................................................................................10 3.1 Panel Session: Steps Toward Regulatory Realization of Digital Twins .......................... 10 3.1.1 Digital Twins - Regulatory Viability ..................................................................11 3.1.2 Xe-100 Licensing Perspectives: Steps Toward Realization of Digital Twins ...................................................................................................11 3.1.3 Using Digital Twins to Support Regulations .....................................................11 3.1.4 Nuclear Energy Institute (NEI) Perspectives on Digital Twins .......................... 11 3.1.5 Kairos Perspective ...........................................................................................12 vii

3.1.6 Westinghouse Perspective ..............................................................................12 APPENDIX A Workshop Attendees......................................................................................A-1 APPENDIX B Presentation Slides ........................................................................................B-1 viii

Page intentionally left blank ix

1 DAY 1 PRESENTATIONS 1.1 Panel Session 1: Industry Applications of Digital Twins in Nuclear In this panel, representatives from the nuclear industry provided a perspective on how digital twins technology has been implemented or may be implemented. The panel provided insights of the potential challenges and benefits of the identified digital twins enabling technologies and introduced additional important enabling technologies. Potential industry strategies for accommodating these technologies within the nuclear lifecycle and any new competencies or technical disciplines needed to support digital twin technologies were discussed as well.

Participants on this session presented the following initiatives/perspectives about the industry applications of digital twins in nuclear:

Xcel Energy discussed their Cap Intelligent Advisor (artificial intelligence (AI) and machine learning (ML)) and Digital Ops Factory programs which facilitates the search, entry, and analysis of data.

The Electric Power Research Institute (EPRI) presented an update of their ongoing project that aims to explore benefits, challenges, and potential use cases for advanced reactors and will establish industry guidelines, best practices, and recommendations for implementing digital twins in advanced reactor life cycle management.

Westinghouse Electric Company discussed practical aspects that can be implemented with the correct use of the digital twin technologies.

Exelon Corporation presented their remote monitoring project. This initiative uses wireless sensors to support plant monitoring.

The Department of Energy (DOE) Advanced Research Projects Agency-Energy (ARPA-E) presented several initiatives that demonstrate that the use of digital twin enabling technologies can decrease operations and maintenance (O&M) labor cost and map temperatures of reactors components, among other benefits.

Metroscope introduced their software that diagnosed equipment problems for all types of plant auxiliary systems.

The presentations slides for Day 1 can be found here and in the Agency Documents Access and Management System (ADAMS) under ML21342A122.

Presentations 1.1.1 Innovation Journey to a Brighter Future Patrick Burke, Vice President of Nuclear Strategy Xcel Energy Presentation Overview: This presentation described several initiatives that Xcel Energy is working on to implement the use of ML/AI to facilitate the search, entry, and analysis of operational data. Other potential efforts were discussed.

1

1.1.2 EPRIs Digital Twin Related Activities for Nuclear Applications Craig Stover, Program Manager Advanced Nuclear Technology EPRI Presentation Overview: During this presentation EPRI discussed in more details the four top innovation areas of interest at this moment. Those are ML and big data, reliable data sharing, digital twinning, and advanced manufacturing. All these areas have great potentials for several possible use cases in the existing and next generation of nuclear power plants.

1.1.3 Westinghouse Perspectives Scott Sidener, Consulting Engineer, Digital Innovation Westinghouse Electric Company Presentation Overview: There was no presentation from Westinghouse. Scott Sidener discussed the companys perspectives on digital twin technologies.

1.1.4 Exelon Nuclear Innovation Projects Overview Rick Szoch and Tim Alvey Exelon Corporation Presentation Overview: Exelon Corporation presented their Innovation Culture and Digital Transformation initiatives. Remote monitoring is their main effort currently, and this project creates a new wireless infrastructure that provides a method to utilize wireless sensors.

1.1.5 ARPA-E GEMINA Portfolio and Digital Twins Charalampos Andreades, Technology to Market Advisor ARPA-E Presentation Overview: ARPA-E introduced the Generating Electricity Managed by Intelligent Nuclear Assets (GEMINA) program and highlighted benefits that digital twins can provide to the industry, including the enabling technologies/capabilities. Dr. Andreades also discussed initiatives that are utilizing digital twin-related technologies.

1.1.6 Digital Twin Monitoring for Advanced Reactors and Plant Modernization Aurélien Schwartz Metroscope Presentation Overview: Metroscope presented their software, which consists of a trusted and reliable digital twin-based program that can diagnose equipment problems for all types of plant auxiliary systems.

2

1.2 Session 2: Advanced Sensors and Instrumentations This session includes presentations focusing on advanced sensors and instrumentation with an emphasis on digital twin applications. The presentations demonstrated the development of new sensors and the greater digital integration of existing sensors. A recurring focus included continuous monitoring control system integration. There is a general goal to move to condition-based monitoring. It was noted that advanced sensors still require testing in the more extreme environments found in advanced reactors. Finally, it was shown how virtual sensors with a digital twin can be used to optimize instrumentation and controls (I&C) systems during the design phase.

Presenters in this session discussed the following use cases of digital twins:

Iterative plant design using digital twins

Digital twin virtual sensors to optimize instrumentation Challenge identified by presenters:

Validation of novel sensor performance under extreme conditions Presentations 1.2.1 Advanced Sensors and Instrumentation for Digital Twin Applications Pattrick Calderoni, National Technical Director, Advanced Sensors and Instrumentation Manager, Measurement Science Department INL Presentation Overview: The development efforts for advanced sensors were outlined.

Advanced sensors include multi-point and multi-modal sensors that must be built to withstand the reactor environment. Advanced sensors will be integrated with control systems to provide real-time data and feedback. Specific sensors were discussed with the predominant feature being high-temperature operation, including in-core neutron flux, high-temperature thermocouples, ultrasonic sensors for temperature measurement and structural health monitoring, and optical fiber-sensors. Wireless technologies and power harvesting were also discussed.

1.2.2 Non-intrusive Temperature and Pressure Wireless Sensor and Transceiver System for Extreme Environment Applications Jorge Carvajal, Fellow Engineer Westinghouse Electric Company Presentation Overview: Ongoing sensor development efforts were reviewed. Wireless sensors for temperature and pressure measurement were presented for use inside sealed fuel rods and inside seal dry storage steel cannisters. The fuel rod sensors would also measure pellet elongation and operate passively. These sensors provide real-time and continuous data. It is suggested that these sensors could be used to increase data availability to accelerate fuel qualification and technical analysis. Efforts are ongoing to test sensors at high temperature and in radiation environments.

3

1.2.3 Data Analytics and Remote Monitoring Integration Molly Strasser Xcel Energy Presentation Overview: Xcel Energy presented their process for moving to a condition-based monitoring and maintenance scheme. Wi-Fi was installed to connect with sensors and smart devices in the hands of personnel. New sensors were added, and existing analogue gauges were integrated into the digital network. Sensing includes vibration, gauge readers, void monitoring, remote radiation mapping, valve position indication, acoustic monitoring for switchgear and transformers, and continuous thermal imaging. Advanced models were used for processing of continuously collected data.

1.2.4 Digital Twin Impact on I&C Systems Development for Xe-100 Matthew Hertel, Senior Nuclear I&C Engineer X-enegy Presentation Overview: The development of I&C systems for Xe-100 was presented. A digital twin was used in lieu of a physical plant. I&C system design was optimized using an iterative process using specifications, transient analysis, and physical systems.

1.2.5 Online Monitoring (OLM) Implementation to Extend Transmitter Calibration Intervals in Nuclear Facilities Brent Shumaker and H.M. Hashemian Analysis and Measurement Services Corporation (AMS)

Presentation Overview: Online monitoring for transmitter calibration was presented. It was presented that transmitters were calibrated at every outage, but typically dont drift in that period. Instead, a condition-based maintenance system using online monitoring was implemented to only calibrate those sensors that had drifted. This system was implemented in Sizewell B since 2005, Vogtle units 1 and 2 since 2018, and the Advanced Test Reactor since 2015.

4

2 DAY 2 PRESENTATIONS

2.1 Session

Use Cases of Digital Twin Enabling Technologies in Nuclear Power Plant Even though digital twin technologies have not been fully implemented in the nuclear industry, they are gaining momentum and support. At this moment there are several projects ongoing to implement this technology into the industry. From risk-informed approaches to optimize equipment maintenance to a specific software that use a combination of digital twin and AI to provide a diagnosis, there are several successful cases in the industry where this technology has been deployed. In this session, participants presented concrete examples and use cases of digital twin technologies in the nuclear industry.

Presenters in this session discussed the following use cases of digital twins:

Preventive maintenance optimization

Work order data analysis

Identification of fault signatures

Development of an automated digital platform

Detection of high-pressure heater leak

Detection of condensate collector tank leak

Detection of condenser losses

Real-time radiation monitoring network (this case is still in the experimental phase)

Challenges identified by presenters:

Lack of synchronization

Lack of good quality data for different fault modes

Data imbalance

Lack of model generalization

Qualification for ML process The presentations slides for Day 2 can be found here and in the ADAMS under ML21342A123.

Presentations 2.1.1 Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Matthew Yarlett, Project Engineer Westinghouse Electric Company Presentation Overview: Presentation of the research results of the project between PKMJ Technical Services, INL, and Public Services Enterprise & Group (PSEG) based on development of condition-based monitoring models and integrating those models into a digital platform for use by the nuclear industry. This project integrated advancements in online monitoring and data analytics techniques with advanced risk assessment methodologies. Preventive maintenance optimization, work order data analysis, and 5

identification of fault signatures, among others, were presented as successful accomplishments of the developed technique.

2.1.2 AI Driven Scalable Condition-based Predictive Maintenance Strategy Koushik A. Manjunatha, PhD Staff Research Scientist INL Presentation Overview: Introduction of a decentralized AI approach using heterogeneous data from a nuclear power plant (NPP) asset to deploy condition-based predictive maintenance strategies. The approach is simple and scalable across different assets at the plant site and across the nuclear fleet. Predictive maintenance models, fault signature identification, fault probability were some examples of results that can be obtained using the technique. This presentation also highlighted several challenges that lead to inaccurate model interpretations.

2.1.3 Thermal Performance Management for Nuclear Power Plant with Digital Twins Christophe Duquennoy, PhD, Nuclear Fleet Thermal Performance Expert

Électricité de France S.A. (EDF)

Presentation Overview: This presentation highlighted use cases on EDF experience. The Metroscope software, which combines a digital twin of the process with AI to perform a diagnosis in operations, is the program utilized by EDF. Early detection of condenser collector tank leaks, condenser loses, and heaters tube rupture are examples of diagnosis activities predicted by the software.

2.1.4 Digital Twin of a Real-time Radiation Monitoring Network Richard McGrath, Principal Technical Leader in Radiation Safety Group EPRI Presentation Overview: This presentation explained the ongoing results of the EPRIs project regarding the use of a digital twin as a real-time radiation monitoring network. This EPRI NextGen RP Project have the potential of optimize the way radiation protection is performed in NPP. The project consists in two phases: Phase 1 - Demonstration of geophysical application for analyzing radiological survey data, and Phase 2 - Apply machine learning to geophysical radiological survey application.

2.1.5 Non-destructive Examination (NDE) 4.0 and ML for In-Service Inspections Iikka Virkkunen, Professor Aalto University Presentation Overview: An introduction to a new system that increases the reliability of the use of NDE for inspections through the application of ML. This new procedure can be integrated to existing NDE methods and allow connected systems to aggregate data. Tools like edge computing and ML are vital for the implementation. The biggest hurdle is the qualification of the system.

6

2.2 Session 2: Digital Twin Enabling Technologies in Advanced Reactors Applications The NRC is preparing to review and regulate a new generation of advanced non-light water reactors, and this session covers the intersection between digital twin enabling technologies and advanced reactor applications. Many companies are planning to use digital twins not only for operation and maintenance, but also in the design, licensing, construction, and decommissioning phases of the NPP lifecycle. Several different approaches to the use of digital twin technologies have been presented, each involving iterations between experiments, simulations, prototypes, and digital twin models. To fully realize the benefits of digital twins, developers need high-quality training data, high-speed surrogate and reduced-order models that can run in real time, and considerable numbers of legacy and advanced sensors to provide the necessary information to the models so that the digital twin can inform predictive maintenance and optimize operational efficiency.

Presenters in this session discussed the following use cases of digital twins:

Anomaly detection with ML

Informing maintenance & security in design space

Detection (fault or anomaly detection), diagnosis (place faults in classes), and health estimation and forecasting (includes performance prediction)

Health evaluation and analysis in real-time, semi-autonomous control room, operational reliability and diagnostics, and safety hazard intervention and limiting defense (detect and mitigate cyber threats)

Recognize and address mechanical and thermal fatigue failure modes which drive O&M activities and costs

Use of DT to fully model all aspects of the system, from physics to controls to properties controlling the device Challenges identified by presenters:

Dealing with the statistically impressive results of ML, which might be individually unreliable

Lack of good quality data for training algorithms and building the DT

Integration of heterogeneous models in a DT

Uncertainty quantification methodology

Getting right balance between digital and physical models in the development process Presentations 2.2.1 Xe-100 Digital Twin Technologies Overview Ian Davis, Senior Digital Twin System Engineer X-energy Presentation Overview: X-energy is using digital twin technologies to support the ongoing development of its advanced high-temperature gas-cooled reactor (HTGR) design and plans to use these technologies to support future operation and maintenance of Xe-100 sites. Some of the inherent characteristics of this advanced reactor concept, such as the robustness of the tri-structural isotropic (TRISO) particles that contain the fuel, low power density of the core, and strong negative temperature coefficient of reactivity, reduce safety concerns and offer 7

opportunities for implementation of new technologies. Some of the digital twin tools include three-dimensional models that can be explored with augmented reality/virtual reality (AR/VR) and coupled to the detailed operator training simulator, a plant historian (includes dashboards and visualizations), and custom ML/AI models for aspects such as conducting predictive maintenance and optimizing operational efficiency.

2.2.2 Humble AI for Reliable Machine Learning-Based Health Twin Dr. Nurali Virani, Lead Scientist General Electric Company (GE)

Presentation Overview: Machine learning-based health twins can be used for fault detection, diagnosis, or health estimations of physical systems and components. A key aspect of using health twins-based automation for critical industrial infrastructure is the development of characterization regions of reliability and trust as safety and performance are paramount. GE has developed an AI program referred to as Humble AI that is aware of its own competence and improves its competence via learning. The crucial element of this program is the model competence evaluation, which analyzes model inputs, model internal representations, and model outputs, and identifies regions of trust, overlap/ambiguity, and extrapolation to get justification-based reliability. The AI was tested using Tennessee Eastman Process simulation data and achieved an overall accuracy of 76% on all the data it was shown; however, on 56%

of the data, the AI program had an accuracy over 99%. This highlights the concept that an AI could be used to automate certain aspects of operation and maintenance (where the confidence and accuracy is high) and request human assistance outside of that region of confidence.

2.2.3 Enabling Technologies for Digital Twins Applications for the KP-FHR Anthonie Cilliers, Senior Management Kairos Power Presentation Overview: Kairos Power plans to use robust digital twin technology in several aspects of the Kairos Power Fluoride Salt-Cooled High-Temperature Reactor (KP-FHR).

Among these are systems for the following functions: safety hazard intervention and event limiting defense (KP-Shield) to provide a passive, robust, reliable safety shutdown capability; operational reliability and diagnostics (KP-Sword) for active plant control; health evaluation and analysis in real-time (KP-Heart) to provide intelligent health monitoring; semi-autonomous industrial grade human machine interface technology (KP-Sight) for a semi-autonomous control room. Kairos power will carry out the development process as follows: 1) small test facilities, 2) large test facilities, 3) prototypical facilities, and 4) commercial facilities. A key aspect of the design is to minimize the safety envelope using state-based plant information.

2.2.4 High-Fidelity Digital Twins for BWRX-300 Critical Systems Emilio Baglietto, Associate Professor of Nuclear Science and Engineering Massachusetts Institute of Technology (MIT)

Presentation Overview: Discussion of the use of high-fidelity digital models using computational fluid dynamics (CFD) to inform maintenance and operational decisions. This is especially useful in new applications that do not have a rich database from which to draw. MIT 8

has been able to demonstrate reduction of operating uncertainty through high-fidelity simulations and accurately predict velocity and temperature fluctuations responsible for fatigue. The STRUCT program developed by MIT was also able to capture complex phenomena driven by the formation and interaction of large turbulent structures that are strongly non-linear and not prone to lumping and generalization. Additionally, MITs program demonstrated accelerations between 50 and 100 times compared to traditional large eddy simulations. These high-fidelity CFD models can be used to create the surrogate models used by a digital twin.

2.2.5 Molten Salt Loop Development Acceleration with Disturbed Single Crystal Harsh Environment Optical Fiber-Sensors Michael Buric, Staff Scientist National Energy Technology Laboratory (NETL)

Presentation Overview: Single crystal optical fibers such as Y3Al5O12 (YAG) or sapphire are used for distributed temperature sensing to map high-radiation and/or high-temperature environments like liquid-fueled molten salt reactors (LFMSRs). The single crystal optical fiber technology can extend into nuclear harsh environments and provide data to not only guide reactor design and improvement through thermal efficiency, but also inform LFMSR transient response. This technology can be used to gather thousands of data points to map reactor coolant temperatures or other parameters, and preliminary testing indicates accuracy with temperatures up to 1000°C and a standoff distance up to 50 feet.

2.2.6 Digital Twin to Production Reactors, The Simulation Continuum Bob Urberger, Chief Software Engineer Radiant Roger Chin, Software Architect Radiant Presentation Overview: Radiant plans on using a digital twin as a common tool between regulators and developers to ensure common sources of information for aspects relevant to them. Throughout the development process, Radiant is advancing their design by using Nuclear Energy Advanced Modeling and Simulation (NEAMS) tools to run high-fidelity multi-physics reactor simulations, and then using the NEAMS results to create reduced- order models for their digital twin software. They iteratively use the high-fidelity NEAMS tools, reduced-order models, hardware-in-the-loop simulations, and subscale or full-scale prototypes to refine both their models and design. Radiant will be able to demonstrate the safety of various operations by using a digital twin run, hybrid simulation run, or full prototype run.

9

3 DAY 3 PRESENTATIONS 3.1 Panel Session: Steps Toward Regulatory Realization of Digital Twins This panel is focused on the intersection between digital twins enabling technologies and regulatory activities. Panelists provided their unique perspectives and insights on how digital twin technology may be employed as a tool for both industry regulatory compliance and perhaps the NRC itself as well as insights into the regulatory outcomes, challenges, resources, and gaps, especially those that are unique or novel, associated with digital twin technologies and areas where regulatory processes should be focused to accommodate these technologies.

Presenters in this session discussed the following use cases of digital twins:

Support visualization with linkage to 3D models with AR/VR, operator simulator training, access to plant historical data, and the development of AI/ML models

Simulator certification, human factor evaluation, and operator workload reduction

Support optimized security staffing via security analysis and what-if security scenarios

Support analysis submitted to NRC for human factor evaluation and staffing

Standardize internal documentation, provide visualizations, and automate analysis

Visualization and creation of virtual sensors to provide greater insights into the actual plant state

Enable efficient and lower risk design with an iterative design process, shifts to virtual design space, and hardware-in-the-loop development and testing

Facilitate the licensing process by structuring documentation and submittal information

Prediction of future operational states using faster-than-real-time simulation

Provide greater operational and regulatory flexibility by calculation of a dynamic operating envelope

Operational anomaly detection

Use of AI/ML for plant control functions, event prediction, equipment remaining useful life estimates, and sensor drift detection

Provide common, rich data for industry and regulators Challenges identified by presenters:

Establishment of an appropriate regulatory guidance for approval of digital twin technologies and applications such as AI/ML control systems, autonomous systems, dynamic operating envelopes, reduced-order and multi-domain models, and reduced plant safety footprints

Implement a streamlined regulatory process and appropriate common information interface to enable rapid regulatory response to plant design changes

Determine appropriate verification, validation, and uncertainty quantification processes for digital twin models The presentations slides for day three can be found here and in the ADAMS under ML21342A124.

10

Presentations 3.1.1 Digital Twins - Regulatory Viability Jeremy Bowen, Deputy Director, Division of Engineering, Office of Nuclear Regulatory Research NRC Presentation Overview: Discussion of NRCs ongoing digital twin research activities which include technical preparedness, regulatory readiness, assessment of standards, and communication and knowledge management; digital twin project completed activities, publications, and takeaways thus far, and active and future project tasks.

3.1.2 Xe-100 Licensing Perspectives: Steps Toward Realization of Digital Twins Tom Braudt, Licensing Engineer X-energy Steve Vaughn, Licensing Engineer X-energy Presentation Overview: Discussion of Xe-100 digital twin tools and their uses for digital control and monitoring, predictive maintenance, high-fidelity simulator development, human factor engineering, operator workload reduction, training, dose reduction, security, fire detection, and to support plant state awareness in different production modes; plans to use DT help analyze operator workload and staffing methodology for topical report to be submitted to the NRC; and plans to use AI/ML for control, event prediction, equipment remaining useful life estimates, and sensor drift detection.

3.1.3 Using Digital Twins to Support Regulations Paul Keutelian, Radiant Presentation Overview: Discussion of Radiants vision to make nuclear portable and use DTs to maximize the speed of design iteration; support regulatory intents to protect personnel, the environment, and hardware, and prove the protections; dynamically assess risk and risk-informed decisions; act as a common source of information for both regulators and developers; and standardize internal documentation, provide visualizations, and automate analysis.

3.1.4 Nuclear Energy Institute (NEI) Perspectives on Digital Twins James Slider, Technical Advisor NEI Presentation Overview: Discussion of NEIs purpose and organization; the importance of a common language for DTs; challenges presented by DT model complexity, real-time inputs, and NRC acceptance and usage for regulatory decisions; NEIs promotion of advanced ideas and best practices within the industry; and NEIs goal to work with industry and the NRC to 11

realize the benefits of DT technologies by establishing a predictable regulatory framework for approving DT applications and protecting public health and safety.

3.1.5 Kairos Perspective Anthonie Cilliers, Senior Manager Instrumentation, Controls and Electrical Kairos Power Presentation Overview: Discussion of the Kairos definition of a DT; DT uses to provide virtual plant sensors, operations databases, support for operator training, and data analytics; demonstrations of smaller subsystems such as a molten salt coolant loop and a virtual counterpart to support a faster iterative design process; how a DT facilitates the intersection between regulatory and design spaces and speeds the licensing process; operational phase use of a DT to predict future plant conditions, create a dynamic operating envelope, and detect anomalies; design phase use to DT to more accurately define safety margins and reduce safety-related footprint within a plant; and use of a DT to reduce design risks.

3.1.6 Westinghouse Perspective Brian Golchert, Principal Engineer Westinghouse Presentation Overview: Discussion of costs associated with DT including lack of guidance for related NRC submissions for techniques such as reduced-order modeling and coupling of single-domain models; the need to establish a DT business case; use of DT to support hardware-in-the-loop; use of DT in place of prototypes to reduce design, development, and regulatory costs; and the need for industry and the NRC to develop guidance needed to implement DTs.

12

Page intentionally left blank 13

APPENDIX A WORKSHOP ATTENDEES First Name Last Name Email Address Mohammad Abdo mohammad.abdo@inl.gov Kamal Abdulraheem kamalabdulraheem@gmail.com Chethan Acharya ckachary@southernco.com Vivek Agarwal vivek.agarwal@inl.gov Indarta Aji indartaaji@gmail.com Ahmed Alshehhi ahmedal565@gmail.com Tim Alvey tim.alvey@exeloncorp.com Harry Andreades charalampos.andreades@hq.doe.gov Michela Angelucci michela.angelucci@phd.unipi.it Kwame Ansah ansahkwame466@gmail.com Todd Anselmi todd.anselmi@inl.gov Thompson Appah appahtompson@gmail.com Jeffrey Arndt arndtjl@westinghouse.com Steven Arndt arndtsa@ornl.gov Dushyant Arora dushyant.arora@siemens.com Oussama Ashy ashyo@ws-corp.com Paridhi Athe pathe@ncsu.edu Md Samdani Azad samdaniazad@konkuk.ac.kr Vittorio Badalassi badalassiv@ornl.gov Jin Whan Bae baej@ornl.gov Emilio Baglietto emiliob@mit.edu Nicholas Baldasaro nick@hoplite.ai Han Bao han.bao@inl.gov Sergiu Basturescu Sergiu.Basturescu@nrc.gov Melissa Bates melissa.bates@nuclear.energy.gov Randall Belles bellesrj@ornl.gov Eric Benner eric.benner@nrc.gov Jacob Benz jacob.benz@pnnl.gov Mounia Berdai mounia.berdai@cnsc-ccsn.gc.ca Satyan Bhongale sbhongale@x-energy.com Harry Bonilla- hbonilla@iastate.edu Tanner Boone tanner.boone@nrc.gov Katarzyna Borowiec borowieck@ornl.gov Jyoti Bose jyoti.bose@alithya.com Jeremy Bowen jeremy.bowen@nrc.gov Thomas Braudt tbraudt@x-energy.com Alexander Brazalovich abrazalovich@x-energy.com Michael Breach michael.breach@nrc.gov Jeren Browning jeren.browning@inl.gov Logan Browning logan.browning@inl.gov John Buchanan jbuchanan@dekabatteries.com Angela Buford Angela.Buford@nrc.gov Michael Buric michael.buric@netl.doe.gov A-1

Pat Burke Patrick.B.Burke@xcelenergy.com Troy Burnett troy.burnett@inl.gov Rob Burns rob@arthur.ai Jonathon Burstein jdburste@bechtel.com Scott Bussey scott.bussey@nrc.gov Dirk Cairns- dirk.cairns-gallimore@nuclear.energy.gov Pattrick Calderoni pattrick.calderoni@inl.gov Clevin Canales ccanales@x-energy.com Salvatore Cancemi salvatore.cancemi@phd.unipi.it Brent Capell bcapell@epri.com Jesse Carlson jesse.carlson@nrc.gov Gene Carpenter gene.carpenter@hq.doe.gov Jorge Carvajal carvajjv@westinghouse.com Arindam Chakraborty achakraborty@viascorp.com Alvin Chan alvin.chan@opg.com Hasan Charkas hcharkas@epri.com Jaydev Chauhan jaydev.chauhan@opg.com Yifeng Che yfche@mit.edu Danny Chien npc1@nrc.gov Roger Chin mroizo40@hotmail.com Roger Chin roger@radiantnuclear.com Helene Chini chinih@westinghouse.com Hangbok Choi Hangbok.Choi@ga.com Anthonie Cilliers cilliers@kairospower.com Stephanie Coffin stephanie.coffin@nrc.gov Christopher Cook christopher.cook@nrc.gov Justin Coury justin.coury@nrc.gov Christopher Crosby ccrosby@osisoft.com Brad Crotts bradley.crotts@orano.group Amy Cubbage amy.cubbage@nrc.gov Samir Darbali samir.darbali@nrc.gov Ian Davis idavis@x-energy.com Niyera Davoodian ndavoodian@gmail.com Grigorios Delipei gkdelipe@ncsu.edu Matt Dennis matthew.dennis@nrc.gov David Desaulniers david.desaulniers@nrc.gov Hadja Fanta Diakhaby nalahadja@gmail.com Xiaoxu Diao diao.38@osu.edu Nam Dinh ntdinh@ncsu.edu Elvis Dominguez dominguezoee@ornl.gov Valentin Drouet valentin.drouet@metroscope.tech Trevor Dudley drtdudley@mozweli.com Christophe Duquennoy christophe.duquennoy@edf.fr Carmen Dykes carmen.dykes@nrc.gov Derek Ebeling-Koning ebelind@westinghouse.com Shannon Eggers shannon.eggers@inl.gov Robert England robert.england@inl.gov A-2

Doug Eskins doug.eskins@nrc.gov Kale Evans kalejevans@gmail.com Nathan Faith Nathan.Faith@ExelonCorp.com Amjad Farah amjad.farah@opg.com Mario Fernandez mario.fernandez@nrc.gov William Ferrell will@ams-corp.com Matthew Ferri mattferri@gmail.com Leo Fifield leo.fifield@pnnl.gov Eric Focht eric.focht@nrc.gov James D Freels freelsjd@gmail.com Raymond Furstenau raymond.furstenau@nrc.gov Pratik Gandhi pratik.gandhi@npxinnovation.ca Alex Garrison alex@radiantnuclear.com Marisol Garrouste mgarrou@umich.edu Ramon Gascot ramon.gascot@nrc.gov Lou Gaussa gaussalw@westinghouse.com Debraj Ghosh dghosh@iisc.ac.in Anders Gilbertson anders.gilbertson@nrc.gov James Godwin drjamesgodwin30@gmail.com Brian Golchert golchebm@westinghouse.com Gregory Golding greggolding@moltexenergy.com Carlos Gonzalez Carlos.Gonzalez@nrc.gov Nicholas Goss nicholas.goss@westinghouse.com Fred Grant ffgrant@sgh.com Scott Greenwood greenwoodms@ornl.gov Donna Guillen Donna.Guillen@inl.gov Anil Gurgen agurgen@ncsu.edu Alexandria Haddad awhadda@sandia.gov Andrew Hahn ashahn@sandia.gov Botros Hanna bn@nmsu.edu Leroy Hardin roy.hardin@gmail.com Leroy Hardin roy.hardin@nrc.gov Brennan Harris brennan.harris@inl.gov Kurt Harris kurt.harris@flibe-energy.com Robert Harwood robert.harwood@slingshotsimulations.co.uk Alex Hashemian alex@ams-corp.com Hash Hashemian hash@ams-corp.com Trey Hathaway Alfred.Hathaway@nrc.gov Gale Hauck hauckge@ornl.gov Jeff Hawkins jeffhawkins@jhawkconsulting.com Robert Hayes rbhayes@ncsu.edu Joe Heit joe.heit@aveva.com Eric Helm eric.helm@framatome.com David Henderson david.henderson@nuclear.energy.gov Peter Henkes phenkes@wisc.edu Richard Henry richard.henry@opg.com Raul Hernandez Raul.hernandez@nrc.gov A-3

Matthew Hertel MHertel@x-energy.com Dan Hoang dan.hoang@nrc.gov Alec Holla alec.holla@npxinnovation.ca Brooks Holland brooks.holland@inl.gov Zachary Hollcraft zachary.hollcraft@nrc.gov Philip Honnold phonnol@sandia.gov Loren Howe loren.howe@nrc.gov Timothy Huddleston timothy.huddleston@inl.gov Nathanael Hudson Nathanael.Hudson@nrc.gov Lauren Hughes lhughes@wpainc.com John Hughey john.hughey@nrc.gov Clyde Huibregtse huibregc@oklo.com Amy Hull amy.hull@nrc.gov Matthew Humberstone Matthew.Humberstone@nrc.gov Eman Ibrahim eman.ibrahim@canada.ca Mesfin Ibrahim mesfin.ibrahim@connect.polyu.hk Dan Isaacs dan@omg.org Raj Iyengar raj.iyengar@nrc.gov Prashant Jain jainpk@ornl.gov Nicholas Jameson nicholas.jameson@inl.gov Patty Jehle Patricia.Jehle@nrc.gov Mike Jenkinson mike.jenkinson@siemens.com Bob Jewart robert.jewart@inl.gov Daniel Ju daniel.ju@nrc.gov Chul Hwan Jung chulhwan.jung@cnsc-ccsn.gc.ca Takanori Kajihara kajihara@tamu.edu Aris Kalafatis aris.kalafatis@opg.com Min-Tsung Kao kaom@ornl.gov Fuad Kassab Junior fuad.kassab@usp.br Maxine Keefe maxine.keefe@nrc.gov Paul Keutelian paul@radiantnuclear.com Genghis Khan khan@ge.com Hamed Khodadadi hakhodadadi1986@gmail.com Anya Kim anya.kim@nrc.gov Paul Klein paul.klein@nrc.gov Brendan Kochunas bkochuna@umich.edu Andrea Kock alk@nrc.gov Alan Konkal alan.konkal@nrc.gov Ben Kosbab bdkosbab@sgh.com Ashish Kotwal ashish.kotwal@und.edu Robert Krawczak krawczrk@westinghouse.com Roman Kuchma wonderwouker@ukr.net Vineet Kumar kumarv@ornl.gov Jonathan Kyle jonathan.kyle@ansys.com Wilson Lam wilsonlam.cns@gmail.com Jeffrey Lane lanejw@zachrynuclear.com John C Lane jcl1@nrc.gov A-4

Kyoung Lee leeko@ornl.gov David Lefrancois david.lefrancois@alithya.com John Lehning jxl4@nrc.gov Matthew Levasseur mplevasseur@bwxt.com Binghui Li binghui.li@inl.gov Jun Liao liaoj@westinghouse.com Bruce Lin bruce.lin@nrc.gov Linyu Lin linyu.lin@inl.gov Yong Chang Liu liuyongchang@gmail.com Deleah Lockridge lockridgedv@ornl.gov Christopher Lohse christopher.lohse@inl.gov Cihang Lu cihanglu@gmail.com Louise Lund Louise.Lund@nrc.gov Lee Maccarone lmaccar@sandia.gov Shah Malik Shah.Malik@nrc.gov Koushik Manjunatha koushik.manjunatha@inl.gov Koushik Manjunatha koush91@gmail.com CS Manohar manohar@iisc.ac.in Jonathan Marcano Jonathan.Marcano@nrc.gov Josh May josh@radiantnuclear.com Richard Mcgrath RMCGRATH@EPRI.COM Noreddine Mesmous noreddine.mesmous@cnsc-ccsn.gc.ca Ernest Mileta Ernest.Mileta@opg.com Jessie Milligan-Taylor Jessica.milligan-taylor@cnsc-ccsn.gc.ca Marwan Mohamed marwan.mohamed@inl.gov Ricardo Moreno ricardo.morenoescudero@ge.com Jawad Moussa Jmoussa@unm.edu Alewyn Mouton alewyn.mouton@opg.com Raheel Naqvi Raheel.naqvi@opg.com Curt Nehrkorn curt.nehrkorn@hq.doe.gov Scott Nelson nelsonsw@ornl.gov Carl Neuschaefer chneusch@aol.com Thien Nguyen thien.duy.ng@gmail.com Daniel Nichols daniel.nichols@nuclear.energy.gov Mirela Nitoi mirela.nitoi@nuclear.ro Kerstun Norman Kerstun.Norman@nrc.gov Alistair Norris alistair.norris@jacobs.com Jesus M Nunez jesus@nuclearalternativeproject.org Bill Obaker obakerwr@westinghouse.com Joe Oncken joseph.oncken@inl.gov Attendee One yadav.vaibhav@gmail.com Ekaterina Paladi Ekaterina.paladi@metroscope.tech Pallavi Pandey pallavip@iisc.ac.in Nithin Panicker panickerns@ornl.gov Sara Perez-Martin sara.perez@kit.edu Eternity Perry eternity@ams-corp.com Angelica Petrovic angelica.petrovic@inl.gov A-5

Jeffrey Poehler jeffrey.poehler@nrc.gov Joseph Poisson jpoisso@entergy.com Cosmin Popescu valentin.popescu@cne.ro Steve Prescott Steven.Prescott@inl.gov Dylan Prevost (Doe) dylan.prevost@nuclear.energy.gov Ivan Price irprice@sandia.gov Craig Primer craig.primer@inl.gov Anthony Qualantone aqualantone@x-energy.com Alexandre Quertamp alexandre.quertamp@metroscope.tech Mihaela Quirk mihaela.quirk@hq.doe.gov Brandon R ricebc@inl.gov Majdi Radaideh radaideh@mit.edu Jean Ragusa jean.ragusa@tamu.edu Pradeep Ramuhalli ramuhallip@ornl.gov Bob Randall bob.randall@nrc.gov Ravi Raveendra rraveendra@cometacoustics.com Wendy Reed wendy.reed@nrc.gov Seyed Reihani sreihani@illinois.edu Mehdi Reisi Fard mehdi.reisifard@nrc.gov Florencia Renteria florenciaren@gmail.com Gustavo Reyes (Inl) gustavo.reyes@inl.gov Alex Rhodes alex.rhodes79@outlook.com Daniel Rosas daniel.rosas@opg.com Cormac Ryan cormac.ryan@aveva.com Will S coinbird@gmail.com Dagistan Sahin dagistan.sahin@nist.gov Osman Sahin osman.celikten@nist.gov Michele Sampson michele.sampson@nrc.gov Erica Sanchez erica.sanchez@inl.gov Daniel Sandoval drsando@sandia.gov Suman Saurav sumanjiseie@gmail.com Abhinac Saxena asaxena@ge.com Abhinav Saxena asaxena@ge.com Randall Schmidt randy.schmidt@exeloncorp.com Paul Schuck paul.schuck@inl.gov Aurelien Schwartz aurelien.schwartz@metroscope.tech Garry Schwarz garry.schwarz@cnsc-ccsn.gc.ca Ting-Leung Sham tingleung.sham@inl.gov Neil Sheehan Neil.Sheehan@NRC.GOV Brent Shumaker brent@ams-corp.com Scott Sidener sidenese@westinghouse.com Paul Sirianni sirianpm@westinghouse.com Alexandra Siwy Alexandra.Siwy@nrc.gov James Slider jes@nei.org Curtis Smith curtis.smith@inl.gov Mohamed Soliman mohamed_saied666666@mail.ru Sharon Soogrim sharon.soogrim@nrc.gov A-6

Christopher Spirito christopher.spirito@inl.gov Antoanela Stoica antoanela.stoica@cne.ro Craig Stover cstover@epri.com Molly Strasser Molly.J.Strasser@xcelenergy.com Cheng Sun cheng.sun@inl.gov Xiaodong Sun xdsun@umich.edu Richard Szoch richard.szoch@exeloncorp.com Emre Tatli tatlie@westinghouse.com Nazila Tehrani nazila.tehrani@nrc.gov Keith Tetter Keith.Tetter@nrc.gov James Tompkins james@radiantnuclear.com Ricardo Torres ricardo.torres@nrc.gov Robert Tregoning robert.tregoning@nrc.gov Panagiotis Tsilifis panagiotis.tsilifis@ge.com Bogdan Tutuianu bogdan.tutuianu@cne.ro Mo Uddin mouddin009@gmail.com Mo Uddin muddin@structint.com Rizwan Uddin rizwan@illinois.edu Christopher Ulmer christopher.ulmer@nrc.gov Troy Unruh troy.unruh@inl.gov Bob Urberger bob@radiantnuclear.com Johannes Van Der Watt johannes.vanderwatt@und.edu Stephen Vaughn svaughn@x-energy.com Justin Vazquez Justin.Vazquez@nrc.gov Swetha Veeraraghavan swethav@iisc.ac.in Rattehalli Vijay rattehalli.vijay@unnpp.gov Purna Vindhya purnavindhya@iisc.ac.in Nurali Virani nurali.virani@ge.com Iikka Virkkunen iikka.virkkunen@aalto.fi Cody Walker cody.walker@inl.gov William Walsh william.walsh@nuclear.energy.gov Congjian Wang congjian.wang@inl.gov Guanyi Wang guanyi.wang@anl.gov Weijun Wang weijun.wang@nrc.gov Justin Weinmeister weinmeistejr@ornl.gov Cindy Wellenbrock ckwellenbrock@gmail.com Timothy West timothy.west@inl.gov Chad Wilhelm cmwilhel@bechtel.com Katherine Wilsdon katherine.wilsdon@inl.gov Paul Witherell paul.witherell@nist.gov Brian Wittick brian.wittick@nrc.gov Jennifer Wong jennifer.wong@opg.com Vaibhav Yadav vaibhav.yadav@inl.gov Xingyue Yang xingyue.yang@inl.gov Matthew Yarlett Matthew.Yarlett@Westinghouse.com Jordan Zenhenko jordan.zenhenko@opg.com Jack Zhao jack.zhao@nrc.gov A-7

Page intentionally left blank APPENDIX B PRESENTATION SLIDES B-1

Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization Ray Furstenau Director, Office of Nuclear Regulatory Research September 14-16, 2021

Future Focused Research (FFR)

FFR program supports the NRC vision of becoming a modern, risk- informed regulator Objectives

Close technical gaps ahead of regulatory needs

Support transformative, innovative ideas

Follow trends across industry and federal agencies

Engage with industry, public, government and university communities

Build new in-house capabilities that will attract and retain top talent Process

Open to ideas from across the agency

Senior Level Staff panel reviews & makes project recommendations

Monitor and communicate progress via program reviews and seminars

Workshop Overview Day 1 Day 2 Day 3 Tuesday, September 14th Wednesday, September 15th Thursday, September 16th 11:00 - 11:00 - 11:00 -

Introduction and Opening Remarks: NRC 11:15 12:45 12:45 Technical Session Panel Session 11:15 -

12:45 Use Cases of Digital Twin Enabling Technologies in Steps Toward Regulatory Realization of Panel Session Nuclear Power Plants Digital Twins Nuclear Industry Applications of Digital Twins 12:45 Break 12:45 Break 12:45 Closing Remarks: NRC 2:00 - 2:00 -

3:45 4:00 Technical Session Technical Session Advanced Sensors & Instrumentations Digital Twin Enabling Technologies in Advanced Reactor Applications 3:45 Adjourn 4:00 Adjourn

Nuclear Industry Applications of Digital Twins Opening Panel Session - September 14, 2021 Moderator: Ray Furstenau, NRC Patrick Burke, Xcel Energy Craig Stover, EPRI Scott Sidener, Westinghouse Rick Szoch & Tim Alvey, Exelon Harry Andreades, ARPA-E Aurélien Schwartz, Metroscope

INNOVATION JOURNEY TO A BRIGHTER ENERGY FUTURE Patrick Burke Vice President Nuclear Strategy September 14, 2021

Leading the Clean Energy Transition A bold vision for a carbon-free future Monticello and Prairie Island Nuclear Plants 38%

Reduction 80% 100%

Reduction Carbon Free 6

Safe, Reliable, and Cost Effective Operation Through Innovation and Technology

Safe - High safety metrics &

ratings (NRC, INPO)

Reliable - Capability Factors 90% to 95%

Cost Effective 33$/Mw to 26$/Mw

Track Record of Innovation Cap Intelligent Advisor (AI/ML) and Digital Ops Factory Search, Entry, Analytics Automation and Work Management integration - GE APM Wi FI and Sensors and Remote Monitoring Early adopters for Organizational Transformation - Fleet Services Model (Operate & Maintain),

Risk based initiatives, TSTF-425 & 505 and 10CFR50.69 Security Innovation and Modification Outage Improvements (Drones, Communications, etc.)

Hydrogen production DOE demonstration project (HTSE)

Flexible Operation - Integration with Wind Exploring operational services with NuScale

Future State High levels of Safe and Sustainable performance through Technology Highly Skilled Multi functional workers that are data and digital fluent Services Organization across multiple units leveraging remote technologies Automation of analytics and reporting Compliance Validation Daily Ops reports - risk based priorities Real time equipment performance monitoring Reporting automation (MSPI, etc)

Integrated operations (Flex, H2, Grid Support, Storage)

Collaboration to Accelerate the Top 4 Innovations EPRI

Contact:

EPRI

Contact:

EPRI

Contact:

EPRI

Contact:

Thiago Seuaciuc-Osorio Rob Austin Hasan Charkas, David Gandy, tseuaciuc-osorio@epri.com raustin@epri.com hcharkas@epri.com dgandy@epri.com 13 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Challenge Who should attend? Attendees will leave with ways to:

EPRI and its co-organizers are

Drive innovation using a new inviting top innovators and network Role Model influencers from around the

Lead a cultural shift in their Courage world: organizations.

Movers

Inspire and on-board others

Shakers into the innovation

Activists movement for responding to

Mirrors climate change, deep

Super-nodes decarbonization, and the Simply healthy restless energy challenge for the people future of nuclear energy.

EPRI Digital Twin Engineering Overview Formed an internal cross-cutting team for collaboration Advanced Developed two technical insights document Modeling and Simulation (3002020014) and (3002022555)

EPRI Digital Twins information video released DT Platform and Integration Monitoring Systems Systems in December 2020 and another one should be released soon.

Near term the team is working on the following: Data Data Analytics Repository What impact do DT applications have on nuclear power plant construction, operation, maintenance and decommissioning? Element of Digital Twins What DT applications can be deployable in the near future?

Digital Twins (DTs) for Advanced Reactors (ARs)

Objective

Explore benefits, challenges and potential use cases for AR applications.

Establish industry guidelines, best practices and recommendations for implementing DTs in ARs life cycle management Status

Summarized use cases for various stages of ARs life cycle

Developed a framework of DT project phases

Selected use cases to further develop DT diagrams for them and to understand needed details to deploy them.

Digital Twin Value Next Steps ?

Summarize experiences and recommendations

Publish in 2022 16 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Digital twins use cases.what are we finding so far?

The skys the limit!. Its important to assess The key question: where does it make sense to use DTs? (Technology readiness, cost benefits, and Technology readiness priorities)

Cost and Value Scalability Regulatory acceptance Applicability Enabling technologies like AI and ML as well as data analytics are important for ensuring successful DT applications 17 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

SCOTT SIDENER CONSULTING ENGINEER, DIGITAL INNOVATION WESTINGHOUSE ELECTRIC COMPANY

Exelon Nuclear Innovation Projects Overview Innovation Culture and Digital Transformation DATA-DRIVEN INSIGHTS HIGH VALUE DECISIONS Extend nExus capability to include digital Data strategy and Big Data Platform Data analytics and AI/ML to achieve value twin and mobile capabilities capabilities through data-driven decision-making Direction for 2022 and beyond DIGITAL PROCESS INNOVATION CONNECTED MOBILE TRANSFORMATION WORKFORCE CAPABILITY Implement process transformations to enable

& CULTURE Expand Use of Smart annual savings Procedures Deliver insights and Innovation Program with the new Future of Work Instructions savings through workflow Constellation automation with AI & ML Innovation Solution Value Capture and Sustainability

Exelon Nuclear Innovation Focus on Value

Competition with natural gas

Aging plants

Plant shutdowns due to economics

Domestic nuclear construction costs

Extension of operating life cycles

Maintaining our domestic nuclear fleet is a matter of national security

Exelon Nuclear Innovation Focus on Value Significant Significant annual annual Potential to benefit savings at full achieve significant implementation annual savings 1000 procedures Several executed dozen so far projects Transform implemented Thousands per site Estimated Large >40,000 Savings hours saved annually by end of 2022

Exelon Nuclear Innovation - Remote Monitoring Wireless Sensors Provides a cost-effective method to enable new innovative ideas Executive Summary:

Problem - The high cost of running wires and using traditional install is cost prohibited. Most data is collected manually by operator rounds Solution -State of art communications infrastructure

Wireless infrastructure that provides a cost-effective method to utilize wireless sensors and enable new innovative ideas and predictive technologies

Low-cost wireless sensors (design once-install many) installed to improve predictive technology and facilitate elimination of high-cost time-based maintenance

Reduce the time that it takes from an idea to implementation New Capabilities: Value:

Innovative Improvement:

There are numerous benefits to the ability to

Wireless sensors are expected to define and transition How did the old system work in terms of people, process, connect Internet of Things devices to the corporate to a future state where maintenance on consequential and tools? network. Ability to utilize a variety of wireless equipment is accomplished "just-in-time". This is Traditional sensors required wires. This process can take years to sensors in the field. The ability to capture and enabled via state-of-the-art diagnostic and analytics design, plan & install. This process can also cost millions of transmit image data back to a centralized server tools, wireless technologies, and an altered work dollars to complete. This process is very slow & costly, and it where it can be used by a variety of applications.

management strategy.

never gets done.

How does the new solution work in terms of people, Application:

process, tools? Other Uses or Potential Enhancements:

Wireless sensors to support plant monitoring included Wireless sensors allow us to rapidly deploy sensors where needed - Maintenance frequency optimization based licensed and unlicensed radio frequency at a much lower cost. These sensors allow us to use design once on procedure results

Cellular devices to support plant communication

& install many concept. Wireless sensors can be installed both - Schedule optimization based on real time

Connecting devices & sensors to Predix APM temporary or permanently installations. They can also be portable task duration or mobile. - Connecting procedures to live plant data

Exelon Nuclear Innovation - Remote Monitoring

Personnel able to perform monitoring from their desk.

Ability for event-triggered actions: Upon an event, predefined actions can be performed, including camera movement and email notification.

Cart setup can be potentially stored in the Power Lab and can be sent for simple assembly and setup at the site with hotspot.

Potential for cost saving by reducing personnel burden and ability to monitor hotspot behavior.

Exelon Nuclear Innovation - Smart Procedures Exelon Nuclear Innovation Analytic Projects Potential Opportunities

Full automation of resource intensive processes

Improved trending and data insight

Enhanced predictive capabilities

Process optimization and forecasting Desired Results

Reduced operating costs

Improved accuracy of decision making

Improve workforce quality of life

More time working on the right stuff

Exelon Nuclear Innovation Analytic Projects Observation Categorization Chemistry Sampling Analytic Initial License Training Throughput Optimization Condition Report Screening Automation

ARPA-E GEMINA Portfolio and Digital Twins Dr. Harry Andreades Technology-to-Market Advisor (support contractor to ARPA-E)

Contact:

Dr. Jenifer Shafer, Program Director, jenifer.shafer@hq.doe.gov December 7, 2021

Defining Digital Twins Mapping of physical asset models in a digital platform where a virtual digital replica is created Consists of three basic building blocks:

1. 3D models

2. Simulators

3. PLM platform to centralize, organize, and manage data Continuous updating (sensors) and real-time data analysis to model physical asset December 7, 2021 ARPA-E GEMINA Portfolio and Digital Twins 30

Digital twins provide a range of benefits Allow for optimal operations and condition-based maintenance Time travel: Allow for manipulation of DT for scenario and what-if analysis without disturbing physical asset (continuously updated data goes beyond static picture)

- This can also apply during design phase - prior to physical asset launch Enables:

Rapid design iterations and optimization

Remote operations

Autonomous power plants

Fleet management

Performance improvement December 7, 2021 ARPA-E GEMINA Portfolio and Digital Twins 31

Where are DTs applied currently?

Oil & Gas, Gas Turbines (and Combined Cycles), Wind Power, Hydro, etc.

December 7, 2021 ARPA-E GEMINA Portfolio and Digital Twins 32

GEMINA (Generating Electricity Managed By Intelligent Nuclear Assets)

Goal: Develop the tools and cost basis for ARs to achieve fixed O&M costs of

$2/MWh without shifting costs to other parts of LCOE Awardee teams are developing the following for one or more of the most promising AR designs:

Digital twins for advanced reactor systems

Relevant cyber physical systems

O&M approaches for advanced reactors

Cost models and design updates December 7, 2021 33

ARPA-E teams are building digital tools for ARs and building blocks for DTs Advanced Reactor Digital Twins Sensors and Data Generation Construction Autonomous Maintenance O&M TEA 34

X-energy: Advanced Operation & Maintenance Techniques Implemented in the Xe-100 Plant Digital Twin to Reduce Fixed O&M Cost 35 35 ADVANCED REACTOR DIGITAL TWINS

GE - Research: AI-ENABLED PREDICTIVE MAINTENANCE DIGITAL TWINS FOR ADVANCED NUCLEAR REACTORS Summary Program Impact AI-enabled predictive maintenance to O&M labor costs BWRX300 from $15/MWh to $3/MWh in an Advanced Nuclear Reactor GE Hitachi Nuclear Program Targets Metric From To Automation Automated workorders Planning staff None labor costs Online calibration Tech and admin staff Predictive Forced outages & trips Maintenance Alarms labor & matl AI-driven predictive algorithms Labor headcount Humble & explainable AI quantify uncertainty to Trust Human establish trust in the models & encourage automation Technology Summary Reactor Operations - Physics-informed machine learning, sensor optimization Reactor Health - Causal, humble & explainable AI for predictive maintenance Decision Making - Autonomous risk-informed decisions for reconfiguration & maintenance 36 ADVANCED REACTOR DIGITAL TWINS

U Michigan: Project SAFARI: Secure Automation for Advanced Reactor Innovation GOAL Reduce NPP O&M costs by delivering a capability which will 4 enable smart functionalities in advanced reactor systems including:

Autonomous, flexible operations Predictive maintenance 3 1 2 Agile Design System and sensors optimization DEMO Kairos FHR END PRODUCT Physics-based, data-enabled, 5 modular and scalable capability that 6 can be extended and applied to any reactor technology 37 37

NETL: DISTRIBUTED MOLTEN SALT LOOP SINGLE-CRYSTAL HARSH-ENVIRONMENT OPTICAL FIBER-SENSORS Introducing fully-distributed sensing to MSRs Growing new cladded single-crystal optical fibers for molten-salt environments Gathering thousands of data-points to map reactor coolant-path temperatures or other parameters Mapping in-core temperature distributions Next-gen sensing replaces single-point sensors like thermocouples First successful growth of Cerium YAG fiber by LHPG Providing data to guide reactor design and improvement through thermal efficiency 38 SENSORS AND DATA GENERATION

SRI: ML FOR AUTOMATED MAINTENANCE OF FUTURE MSR Create an integrated software and algorithm architecture to teach automated systems from simulated data Why - Autonomous maintenance operations identified as critical for next generation MSR Component creation, Establish baseline, vendor agnostic methodologies manipulation and Look to solve two (2) problem sets reality - feedback

1. Robot positioning AWS

2. Tracking via reinforcement learning (RL)

IRTC SR Virtual Construct Physical demonstration 1 DEFT & SR Algorithm development ORNL Physical Physical demonstration 2 Robot Physical Virtual Physical maintenance Virtual equipment replicates Machine Learning for Automated Maintenance of Future MSR 39 AUTONOMOUS MAINTENANCE

NSCU: A Data-driven Approach to High Precision Construction and Reduced Overnight Cost and Schedule Reality Capture Integrated Project (drone+laser scanner) 4D BIM Site Model CONSTRUCTION 40

EPRI: BUILD-TO-REPLACE: A NEW PARADIGM FOR REDUCING ADVANCED REACTOR O&M COSTS Identify representative SSCs for evaluation of design life implications for cost and performance associated with licensing, construction, operation, maintenance, and decommissioning Identify at least two reference advanced reactor (AR) designs to establish baseline O&M cost, enveloping multiple missions and a broad plant parameter envelope

- Target: small modular light-water reactor and high-temperature gas-cooled reactor as mature technologies,

- Aspirational goal: extend to a molten salt reactor design Define scenarios for reduced SSC lifetimes to evaluate impacts on O&M costs against other categories, including construction and decommissioning Metric State of the Art Proposed Fixed O&M cost $20/MWh for light-water reactor (LWR) fleet < $10/MWh for more mature ARs

< $5/MWh for less mature ARs Design life for major 30 - 40 years for PWR steam generator Major SSCs < 15 years plant components 60+ years for RPV No life limiting SSCs 41 O&M TEA

SUMMARY

ARPA-E has a complimentary fission portfolio which targets both capital and operating costs reductions for making advanced reactors commercially competitive Digital twins and their enabling technology can be disruptive in changing the design, construction, and operating phases for advanced reactors Multidisciplinary capabilities are needed for successful implementation of DTs ARPA-E has a strong technology-to-market focus and encourages/enables performers to focus on commercial relevance and commercialization aspects of their technology Performer material is from unrestricted/publicly-available info from the annual ARPA-E Nuclear Program Review meeting, and can be found at https://arpa-e.energy.gov/2021-annual-nuclear-review-meeting.

ARPA-E GEMINA Portfolio and Digital Twins 42

Digital Twin Monitoring for Advanced Reactors and Plant Modernization Webinar - 09/14/2021 Aurélien Schwartz Software company founded in Paris in 2018, with offices in Germany and the USA Patented technology originating from the R&D Center of EDF constituted of more than 2000 researchers

Starting with a usecase Tube rupture in a High Pressure Feedwater Reheater (Blayais, France)

Automatically detected on Dec 23 with a magnitude of 3kg/s Fixed at the end of Feb with a magnitude of the leak of 15 kg/s

Key principles Sensors First, we use a model to build live symptoms for the plant Expected behavior Digital Twin Symptoms

=

We then use a failure library embedded in the Failure library digital twin to preform a root cause analysis.

Inferential Engine

= )

Failures are classified by magnitude, impact Diagnostics and likelihood.

3

Physical based model The model of a NPP is composed of around 12 000 equations and variable declarations.

It simulates both the nominal behaviour and failures of the plant

Inferential engine The inferential engine performs a root causes analysis of underlying problems, through the analysis of the symptoms. It solves the following probability : P (failure 1, failure 2, failure p l Symptoms)

The software Software UX designed with operating teams Reliability of 90% observed in 2020 - 2000 GWh energy loss detected - Over 300 users

Technological landscape Monitoring of power plants aims at understanding any problem the plant during the operations Soon we think that Physical Models and ML approaches will feed a unique diagnostic engine, making the best out of available process data Explore Detect Understand KPIs Physical models Root cause identification Production ready Synoptics Supervised ML Under Statistics Unsupervised Machine development Learning (APR)

Data viz

Innovation with ARPA-E Asset Performance and Reliability for the HTGR Reactor Cavity Cooling System Using Metroscope If we change the digital twin, we change the diagnostics

In a nutshell Diagnosis Digital Twin Approach for HTGR - RCCS at Varying Power Opportunity: Create trusted, reliable Digital Twin-based diagnoses of equipment System P&ID and Process Data Process calibration data design information problems for all types of plant auxiliary systems - beginning with advanced Filtered Measurements reactors Digital Twin Metroscope Success: Achieve a generalized approach Topological Nominal Model Expected Values Symptom Model Detection to digital twins that can anticipate new faults over a variety of system modes and can be deployed in a commercial offer with Metroscope Modeled Faults Failure Library Diagnosis demonstrated ROI Algorithm

Contributors Builds the digital twin and preforms the research work Eric Helm Mary Beth Principal Investigator Baker Project Manager Provides diagnostics technology and software platform Provides cyber-digital asset test data and technical consultation Todd Matthews Pascal Brocheny TH Modeling and TH Modeling Application Developer

Other current focus Tomorrow Christophe Duquennoy will be speaking about diversification on cooling tower and return of experience at EDF Digital twin are already ready for Gas Plants, Diesels Plants but also common industrial assets such as Data Center cooling processes Over 50GW and 60 industrial units equipped worldwide

Thank you for your attention Advanced Sensors and Instrumentations Technical Session - September 14, 2021 Moderator: Hasan Charkas, EPRI Pattrick Calderoni, INL Jorge Carvajal, Westinghouse Molly Strasser, Xcel Energy Matt Hertel, X-energy Hash Hashemian & Brent Shumaker, AMS

February 10, 2021 Pattrick Calderoni - INL National Technical Director, Advanced Sensors and Instrumentation Manager, Measurement Science Department Advanced Sensors and Instrumentation for Digital Twin applications Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 14, 2021

DOE Advanced Sensors and Instrumentation program Mission Vision Address critical technology gaps for monitoring and controlling existing ASI research results in advanced sensors and I&C technologies that are and advanced reactors and supporting fuel cycle development qualified, validated, and ready to be adopted by the nuclear industry

Program objectives FY22-25

Sensors for advanced reactors Develop advanced sensors (multi-mode; multi-point/distributed; miniature size and limited or no penetrations) and supporting technology (rad-hard electronics, wireless communication, power harvesting) for nuclear instrumentation Demonstrate nuclear instrumentation performance in conditions relevant to advanced reactors (including irradiation)

Establish a supply chain for advanced reactor instrumentation (fabrication and services)

Instrumentation for irradiation experiments Provide real time instrumentation and passive monitors to measure local operational parameters (neutron flux, temperature, pressure, mechanical solicitations) in TREAT, ATR, HFIR and MITR experiments Develop methods to characterize nuclear fuel and material properties (thermal conductivity, microstructure, mechanical behavior) during irradiation Instrumented capsule for LWR fuel safety test in TREAT In-core flux Beryllium Booster fuel Neutron flux sensors (SPND, fission chambers sensors plug element and dosimetry) deployed in ATRC to characterize the I-loop booster performance

Program objectives FY22-25 A logical progression towards sensor-based

Digital Technology for autonomous operation of advanced reactors advanced reactors Integrate advanced sensors and instrumentation in Nuclear Digital Twins (NDT) with Hardware in the Loop simulation for the AUTOMATED CONTROL PERFORMANCE MONITORING phased demonstration of Supervisory algorithm Physics-based pattern recognition performance-based control algorithms to enable autonomous operation Develop condition monitoring technologies for anomaly OFF-NORMAL EQUIPMENT HEALTH MONITORING DECISION MAKING Automated reasoning diagnostics detection, diagnostics, Reinforcement Learning prognostics, and decision making that can operate on streaming data Develop modeling and simulation tools for MAINTENANCE SHEDULING communication technologies to Markov Process optimization support integration with control systems

Sensing modalities In-core Multi-point and distributed measurements of process variable fields

- Core-wide estimation of temperature, power, approach to safety limits etc.

Optical fiber

Sensing modalities In-vessel Imaging opaque environments for in-service inspection Pulser Receiver Gain Filter DAQ & Image Gas Flow Processing Control Center Sodium Flow Control Center

Sensing modalities Ex-vessel Plant-wide sensing combined with process models for full two-phase mixture two-phase mixture from HP turbine from MSRs ES-PT-114A ES- PT-100A -

D1 D3 FE-105A state awareness TE-111A PT-100A

--- VENT ---

"A" SG Feedwater Inlet VENT L1

--- 1-CN-PT-150A Drains from MSR 1A-1-CN-TE-140A 1B-1C-1D Shell-side FE-105B -

TE-111B PT-100B FW-PT-158 I

"B" SG Feedwater Inlet FW-TE-110A FW-TE-109A 1-CN-PT-150B 1-CN-TE-139A 1-CN-TE-137A L2 F1 E1 1-CN-TE-140B F3 E3 1-SD-FT-100C 1-SD-PT-108C 1-SD-LT-111C FE-105C SD-FT-102A TE-111C SD-TE-110A G1 1-SD-TE-111C ---

1-SD-WY-100C

--- 1-CN-PT-150C 1-SD-PT-100C SD-TE-109A PT-100C 1-CN-TE-140C "C" SG Feedwater Inlet - -

- 1-SD-FT-100B 1-SD-PT-108B 1-SD-WY-100B L3 ES- PT-100B 1-SD-TE-111B ---

D2 1-SD-PT-100B 1-SD-PT-100C

- 1-SD-FT-100A 1-SD-PT-108A 1-SD-WY-100A VENT 1-SD-TE-111A --- 1-SD-PT-100A 1-SD-FT-100C -

1-SD-LT-111A 1-SD-PT-108C 1-CN-PT-101B

- - G3 1-SD-WY-100C -

1-SD-TE-111C --- FW-TE-110B FW-TE-109B 1-SD-PT-100A 1-SD-PT-100C F2 --- -

E2

- SD-FT-102B G2 1-SD-FT-100B SD-TE-110B ES-PT-114B D4 1-SD-PT-108B 1-SD-WY-100B -

VENT 1-SD-TE-111B --- 1-SD-PT-100B - ---

SD-FT-100A 1-SD-PT-108A 1-SD-WY-100A 1-CN-TE-139B 1-CN-TE-137B 1-SD-TE-111A --- 1-SD-PT-100A 1-SD-LT-111B F4 E4 SD-TE-109B

- G4

- 1-SD-PT-100B

Technology demonstration - in-core neutron flux sensors ATR-C test objectives:

Testing instruments in representative environments (SPND, FC, dosimeters),

Neutron and gamma detection

Fast (Hf, Gd) and slow (Rh, Vd) time

Developing key domestic expertise for in-response core instrumentation,

Established design and fabrication process at INL

Supporting characterization of test positions

Calibration and temperature compensation development in NRAD (Iloop booster)

(summer 2021)

Performance demonstration in TREAT, AGR5/6/7, ATRC, and MITR (FY22)

Test rig was installed in ATRC I-13 position on Self Power 2/11/2021 and irradiated for six hours.

Detectors

VTR project on fast spectrum SPND - design optimization using ORNL Geant4 code for Ta emitter

NRAD test include fission chambers from CEA (Loic Barbot 2 months visit) and Photonis TREAT pulse transient with Gd- and Hf-SPNDs compared to an ex-core detector.

Technology demonstration - TCs for fuel and cladding materials Table 1: Summary of performance parameters for the HTIR-TC Performance Performance Requirement Performance Parameter Fuel Test Application Requirement Stand-Alone Application

Mo-Nb junction for high temperature applications (1600 C) and low drift Temperature Room Temperature - 1600°C Room Temperature -

under neutron irradiation Range 1600°C

Performance demonstration in AGR5/6/7 - highest temperature ever Accuracy Not Specified +/-1%

recorded in pile without drift (1482 C) Drift 3% for 4.5 x 10 nvt 21 3% for 4.5 x 1021 nvt

Design optimization: corrosive environments, multi-point detection (thermal) (thermal)

Commercialization: TCF with Idaho Life 4.5 x 10 nvt (thermal), or 10 18 months or 4.5 x 1021 21 Labs Corp, ASTM standard and industrial qualification at AMS thermal shocks (room nvt (thermal) temperature to 1600°C)

High Temperature Irradiation Mechanical Resistant (HTIR) thermocouple Ruggedness:

Rugged Junction Rugged mechanical junction Rugged mechanical design junction design Bend Radius Minimum of 0.5 inch Minimum of 0.5 inch Thermal Shock 5 sudden startups and 5 100°C/hr sudden shutdownseach causing a thermal shock on the order of room temperature up to 1600°C Response Time <0.5 seconds <0.5 seconds HTIR response compared with standard types

Technology demonstration - acoustic sensors UT operational window defined in high temperature furnace (up to 2200 C):

Sheathed multi-waveguide materials: 316-stainless steel, lanthanated molybdenum, and zircaloy-4

Non sheathed single waveguide materials: 316-stainless steel, molybdenum and tungsten

Ultrasound based sensors enable distributed temperature measurements up to 2200ºC

INL had demonstrated the reliability of magnetostrictive material transducers under irradiation

Current research focuses on waveguide design optimization and unfolding signal response of distributed measurements Ultrasonic Thermometers

UT performance modeling and design optimization is ongoing

UT demonstration continues in irradiation experiments (DISECT in BR2, TREAT, MITR)

Consolidating work on Surface Acoustic Wave sensors development based on radiation resistance materials (AlN, LiNbO)

Technology demonstration - optical fiber sensors Distributed Temperature Sensing (DTS) for DRIFT experiment in

Advanced sensor configuration and TREAT: temperature profile interrogation techniques to measure: along the length of a single fiber

Distributed temperature, strain and vibration

Fission gas pressure and composition

Engineering solutions for sensor packaging, pressure feeds

Active compensation techniques for OF sensors operating in radiation environments Optical Spatially resolved time Fibers dependance:

Black traces are radially closer to fuel

Excellent symmetry

Effects of heat sink become more important after 1 min Prototype optical fiber pressure sensor Optical Frequency Domain based on Fabry-Perot Reflectometry (OFDR) for interferometry temperature mapping of TREAT heat sink

Technology development

Ultrasonic transducers and sensor

High-temperature & high-power-

Automated technology coupled with

Passive wireless sensor technology systems operating in extreme density capability for in-core or in- advanced data analytics for based on a network of digitally conditions with long operational life vessel power harvesting assessing the health of pipes in printed radio frequency (RF) surface

The use of Z-cut LiNbO3

Technology development through nuclear power plants as the pipe acoustic wave (SAW) sensors piezoelements coated with Cr/Au ASI funded project at the University material degrades due to corrosion

Enable multi-point and multi-mode thin film allows high temperature of Notre Dame (Yanliang Zhang)

Combines innovations in materials sensing (temperature, hydrogen gas, operation (800 C) and rad tolerant

Performance demonstration in MITR for sensing both chemical and voltage, and current)

Broadband transducer (bandwidth up though NSUF funded project mechanical degradation with

Demonstration included two-antenna to 30 MHz) now being coupled with statistical algorithms based on relay for communication / sensing temperature, pressure, flow and Bayesian modeling. through RF-opaque materials SHM acoustic emission sensors Radiation Endurance Ultrasonic Thermoelectric generators (TEGs) Diagnostics and Prognostics of Direct Digital Printing of Passive Transducer, X-wave Innovation Inc for power harvesting Corrosion Processes in Pipes Wireless Sensors Thermocouples Material removal on installed on inside of elbow outside diameter REUT prototype, (left) Version 1, (right) Version 2, and (up) CAD model of Version 3 Printed SAWs developed and fabricated at ORNL Courtesy of Uday Singh and Dan Xiang,

- courtesy of Tim McIntyre SBIR grants recipients D. Adams, Vanderbilt University Y. Zhang, University of Notre Dame V. Agarwal, Idaho National Laboratory

Non-Intrusive Temperature and Pressure Wireless Sensor and Transceiver System for Extreme Environment Applications Jorge Carvajal, Fellow Engineer September 2021 Presented at Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization workshop 69

Non-intrusive real-time data

Both Commercial & DOE canisters

Applications requiring wireless real- time

Accelerates new fuel development

Accelerates licensing of casks surveillance in extreme conditions

Improved efficiency of interim storage

Nuclear safeguards

Accelerates new fuel licensing

Develop basis for dry storage of new fuels

Increase operating margins at plants

Nuclear defense

Utility fuel handling no longer reliant on

Enhances instrumentation for DOE test reactors conservative models:

Gas pipelines

Fuel can be removed from pool earlier

Gas storage

Reduced drying cycles

Increased heat load margins in storage

Etc.

Enables canister surveillance monitoring

Wireless sensor located in a fuel rod provides real-time data

centerline fuel pellet temperature

pellet elongation

rod internal pressure

Facilitates licensing of new fuel products

Enhances plant operation through improved utilization of margins

Similar to in-rod data collection techniques used in test reactors such as Halden but with wireless data transmission

Delta in the coupling amplitude is proportional to measurement of interest

U.S. patent application serial No. 16/214445 and 16/564150

Westinghouse Non-Proprietary Class 3 © 2021 Westinghouse Electric Company LLC. All Rights Reserved. 72 Benefits Ideal Timeline for Current Approach to Development and Licensing of New Nuclear Technology

Non-intrusive real-time data Years 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Cool instead of the typical cook and Design, build and qualify for down Data evaluation Test reactor testing and PIE testing new and model look significantly accelerates technology trans-port generation development Data Flow Cool Data

New approach significantly Commercial reactor down and PIE evaluation NRC review and and NRC final approval testing trans-reduces time for licensing, port submittal increases reliability of models and Ideal Timeline for Improved Approach to Development and Licensing of New Nuclear Technology therefore licensing decisions Years 0 1 2 3 4 5 6 7 8 9 10 11 Commercial Design, build reactor

Significant enhancement to in-core and qualify for testing new Test reactor testing with in-rod sensors testing with in-rod technology instrumentation for test reactors, sensors Data Flow especially National Laboratory Data evaluation and model validation Reactors (e.g. INL ATR and ORNL and NRC submital updates HFIR) Atomic Data Flow Data models evaluati NRC review to on and NRC review and NRC and final predict preliminary approval submit- approval fuel behavior tal

ORNL HFIR Irradiation & PIE (NSUF project)

- Temperature & Pressure sensor to be installed in HFIR late 2021

- Expected temperature range 300°C - 500°C

- 5 cycles, total dose ~3X1021 n/cm2

- Fuel surrogate simulates temp source

- External gas injection simulates rod internal pressure

- Continuous data collection

WEC Long Term Temperature test

- Two sensors to be installed in autoclave at prototypical fuel rod pressures

- Held at ~400°C for 6 - 12 months

- Continuous data collection P. Mulligan, In-Core Neutron Flux, Temperature, And Pressure Instrumentation For The Wire-21 Experiment In The High Flux Isotope Reactor, 12th International Conference On Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies (NPIC & HMIT), June 2021.

Technology derived from the Westinghouse Accident Tolerant Fuel (ATF) In-rod sensor development

Wireless sensor located inside dry cask steel canister provides real time data such as temperature and pressure

Sensor does not require penetrations to the canister

Multiple sensors can be interrogated simultaneously

Sensor lifetime > 40 years (based on dry cask 3.5 Grad 40-year TID estimate).

Long term maximum operating temperature 425C S represents the sensors and TR

U.S. patent application serial No. 16/448706 represents the transceiver

Online data potentially enables the reduction in time and cost of spent fuel management.

Utilities are no longer reliant on overly conservative peak clad temperature (PCT) models, which prescribe:

- When fuel is removed from spent fuel pool

- Heat load margins in dry storage canisters

- Drying cycle duration

- Canister susceptibility to degradation (e.g. chloride induced stress corrosion cracking, etc.)

Accelerates dry storage and transportation of advanced fuels

- NRC will require data to develop technical basis of dry storage of new materials

- Synergy with high enrichment, high burnup, and ATF

Dry cask inventory at utility is growing given the lack of a permanent repository location

Accelerate development and qualification of remote wireless sensors for use in extreme environments, e.g. high radiation fields and temperatures

Phase I: Establish Design Criteria

- Develop Sensors Performance Requirements Document

- Identify Key Objectives, Functional Criteria, Survivability Needs, etc.

Separate Dry Storage, Transportation, and Disposal Considerations

Phase II: Laboratory Scale Testing and Qualifications

Phase III:Full Scale Testing and Validation

- Canister Mockups and Loaded Canister Demonstrations

Power cycling & temperature measurement results Neutron flux = 1e14 n/cm2/sec Coolant temperature = 300°C Component max temperature (gamma heating) ~ 500°C Ceramic enclosure housing sensor components SS enclosure holds ceramic enclosure within autoclave PWR loop MITR Core with water loop position open Sensor frequency accurately tracks internal thermocouple and reactor power

Two resonant sensor circuits assembled in a metallic enclosures and remotely interrogated

Received signal cannot be resolved in time domain. FFT necessary to resolve signal Multiple sensors in the frequency (left) and time (right) domain Sensors at two unique frequencies with remote interrogator

Westinghouse Non-Proprietary Class 3 © 2021 Westinghouse Electric Company LLC. All Rights Reserved. 79 Temperature & Gamma Irradiation Test Blue - TC Purple - Sensor High temperature oven & data acquisition system

-Minimal to no change in inductance values Sensor signal vs. pressure

-Slight increase in inductors series resistance Gamma testing: Clean Hot Cell exterior (left) and interior (right)

Thank you!

Jorge Carvajal, carvajjv@westinghouse.com 80

Data Analytics and Remote Monitoring Integration Molly Straser

Xcel Energy Company Overview

3.7 million electric customers

2.1 million natural gas customers

Monitoring & Diagnostics (M&D)

Utilizing advanced analytics to proactively identify equipment risk, leading to reduced maintenance expense and unplanned outage impact, while enabling condition-based maintenance practices and operational excellence.

Advanced Pattern Recognition (APR) predictive analytics identify statistical deviations of modeled operating parameters.

Xcel Energy Monitoring &

Diagnostics (M&D) 2014: Inception - 7 plant pilot program 2016: Major thermal units included:14 plants, 35 units 2017: Expansion to wind, 3 nuclear units 2020: 13,135 MW thermal generation

846 MW Wind generation, 2000 MW Wind EOY 2020 Predictive Analytics - Mechanical condition monitoring

Failure modes limited by plant instrumentation

$29M avoided and hard savings through Q1 2020

>4000 actionable advisories

M&D Success & Results Operational Excellence Avoided Cost Examples

Excessive desuperheat spray leading to

Wind turbine gearbox - numerous early gear HRSG tube damage defects avoiding gearbox replacement ~ $350k

Feedwater regulator closed resulting per event in low fuel gas temp - damage

Air heater guide bearing temperature increase, prevented lube oil supply problem corrected

Excessive feedwater heater draining -

Steam turbine vibration changes, balancing prior low levels, erratic levels to forced event

Poor condenser performance,

Fan bearing temperature increase, cooler efficiency impact operation corrected preventing failure

Boiler acoustic leak indication, operation mitigation until scheduled outage Direct Savings Examples

Major maintenance deferrals, known good condition and performance, i.e., BFP overhaul elimination ~ $250k

Capital budget reduction for wind turbine gearbox replacements, early fault identification and known condition

Condition based maintenance - known good condition allows for delay or elimination of scheduled or calendar-based maintenance - expansion priority

Xcel Energy Nuclear Innovation: Sensor Infrastructure Mechanical Sensors

Vibration Sensors

Wireless Gauge Readers

Void Monitoring

Remote Radiation Mapping

Valve Position Indication Electrical Sensors

EPRI Acoustic Monitors for Transformers

EPRI Disconnect Switch Monitor

Continuous Thermal Imaging Wi-Fi Devices

AR Headsets / iPhones / Tablets

Xcel Energy Nuclear Innovation:

USA Advanced Remote Monitoring

Xcel is working with INL to begin development of a method to streamline current pain points in the M&D architecture

Part of a larger initiative with collaboration with USA plants, as well as Idaho National Lab

Standardized Monitoring and Diagnostics (M&D) Software Platform

Automatic thermal performance and fire detection using image/video recognition tied into cyber compliant systems

Beginning to automate operator round data collection

Transformer and cycle isolation monitoring

Begins to apply machine learning to monitoring limits (Xcel collaboratively with INL)

Questions?

Molly Strasser Nuclear Innovation Manager Molly.J.Strasser@xcelenergy.com

2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization Advanced Sensors and Instrumentation Digital Twin Impact on I&C Systems Development for Xe-100 Matthew Hertel, Senior Nuclear I&C Engineer September 14th, 2021

UCO TRISO Particle - Primary Fission Product Barrier 0.855 mm 19 000 0.425 mm TRISO coated 60 220 000 UCO kernel particles in a mm pebbles in Porous Carbon the core pebble Pyrolytic Carbon Silicon Carbide Pyrolytic Carbon Primary safety goal is to ensure that fission products are retained within the TRISO coated fuel particles to the maximum extent possible This is achieved through production of high quality TRISO fuel and ensuring that temperatures in the core never exceed the temperatures for which the fuel has been tested (AGR Experiments)

Background:

Xe-100 Plant Overview Standard X-energy plant have 4 Reactors

- 4 Turbines producing 320 MWe, attributes include:

200MWth/80MWe Per Module Process heat applications Proven intrinsically safe Meltdown proof Walk-away safe Modular construction Requires less time to construct (2.5-4 years)

Road transportable for diverse geographic areas Uses factory-produced components Load-following to 40% power within 15 minutes Continuous fueling; resilient on-site fuel storage

Digital Twin I&C Development Cycle I&C Systems Design Input Initial Model Design Start Here Process Model ???

Plant Model

I&C Systems Engineering Process Flow + Digital Twin Toolsets Early Stage

+ +

Large Changes P&ID and Design Specifications Transient Analysis Physics Simulation Transition Concept Requirements System System Design System Test &

Operation &

Development Engineering Architecture & Development Integration Evaluation Maintenance Late Stage Plant AI / ML Historian + Models Small Changes

DT Enabled Design by Analysis: IPS Trip Setpoints Select Initial Steady Simulate Select IPS Initial Plant Model State Open Loop Sensors &

IPS (Digital Twin) Simulation Transient Actuators Setpoints Improved Simulate Evaluate Yes Approve Within Plant Model Protected Plant IPS Margin?

(Digital Twin) Transient Response Setpoints Adjust No IPS Setpoints External Development Cycle

DT Enabled Design by Analysis: Control Trade Study Common Controlled Variable Manipulated Variable Secondary Variable Feedwater Valve dP Feedwater Pump Speed Feedwater Valve Flow Deaerator Level CIFW Valve Position CIFW Flow Deaerator Pressure HPT Valve Position HPT Valve Flow FWP Recirculation Flow FWP Recirculation Valve Position Flow/Speed Ratio Turbine Header Pressure Turbine Bypass Valve N/A Option 1 Controlled Variable Manipulated Variable Secondary Variable Main Steam Pressure Feedwater Valve Position Feedwater Flow Reactor Outlet Temperature Control Rod Depth Reactor Power Main Steam Temperature Circulator Speed Helium Mass Flow Turbine Power Turbine Throttle Valve Position Steam Flow Option 2 Controlled Variable Manipulated Variable Secondary Variable Main Steam Pressure Circulator Speed Helium Mass Flow Reactor Outlet Temperature Control Rod Depth Reactor Power Main Steam Temperature Feedwater Valve Position Feedwater Flow Turbine Power Turbine Throttle Valve Position Steam Flow Option 3 Controlled Variable Manipulated Variable Secondary Variable Main Steam Pressure Feedwater Valve Position Feedwater Flow Reactor Inlet Temperature Circulator Speed Helium Mass Flow Main Steam Temperature Control Rod Depth Reactor Power Turbine Power Turbine Throttle Valve Position Steam Flow Option 4 (THTR)

Controlled Variable Manipulated Variable Secondary Variable Main Steam Pressure Circulator Speed Helium Mass Flow Reactor Inlet Temperature Feedwater Valve Position Feedwater Flow Main Steam Temperature Control Rod Depth Reactor Power Turbine Power Turbine Throttle Valve Position Steam Flow

DT Enabled Design by Analysis: Xe-100 Next Steps Controls Design Instrument Design

Online Monitoring (OLM)

Implementation to Extend Transmitter Calibration Intervals in Nuclear Facilities Presented by:

Brent Shumaker and H.M. Hashemian AMS Corporation Presented for:

2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 2021

All Plants Calibrate Transmitters Every Cycle Procedure Prepare M&TE Dress out for containment entry Remove channel from service Valve-off sensing lines Inject test signal to transmitter Calibrate (if needed)

Return everything to service Drawbacks Radiation exposure Potential to damage transmitters Up to 5% experience maintenance-induced errors Farley recently replaced numerous manifold valves due to wear and tear Increased outage maintenance and critical path time September 2021 www.ams-corp.com SLIDE 100 OF 14

Why Do We Say Transmitters Drift Very Little?

(10-year calibration history of a typical nuclear grade transmitter)

+ As Found Limit

- As Found Limit September 2021 www.ams-corp.com SLIDE 101 OF 14

Online Monitoring (OLM) Identifies Drifting Transmitters (Actual PWR Plant Data)

No Drift Deviation 0 1 2 Time (Years)

Drift Deviation 0 7 14 Time (Months)

September 2021 www.ams-corp.com SLIDE 102 OF 14

Traditional Calibration vs. OLM Traditional Calibration OLM Step 1. Determine if Step 1. Determine if calibration is needed calibration is needed Step 2. Calibrate if needed Step 2. Calibrate if needed September 2021 www.ams-corp.com SLIDE 103 OF 14

Typical OLM Results Item Group Name Tag Name Result 1 SG C OUTLET PRESSURE PT0494 Good 2 SG C OUTLET PRESSURE PT0495 Good 3 SG C OUTLET PRESSURE PT0496 Good 4 PRESSURIZER LEVEL LT0459 Good 5 PRESSURIZER LEVEL LT0460 Good 6 PRESSURIZER LEVEL LT0461 Good 7 PRESSURIZER PRESSURE PT0455 Bad 8 PRESSURIZER PRESSURE PT0456 Good 9 PRESSURIZER PRESSURE PT0457 Good 10 PRESSURIZER PRESSURE PT0444A Good 11 PRESSURIZER PRESSURE PT0445A Good September 2021 www.ams-corp.com SLIDE 104 OF 14

OLM Process for Pressure Transmitter Calibration Extension

Retrieve data from plant computer

Analyze data to identify transmitters that have drifted out of tolerance

Provide a list of transmitters to be calibrated Item Group Name Tag Name Result Startup Steady State Shutdown SG A OUTLET Data Data Data 1 PRESSURE PT0474 Good SG A OUTLET 2 PRESSURE PT0475 Good Transmitter Reading SG A OUTLET 3 PRESSURE PT0476 Good SG A NARROW 4 RANGE LEVEL LT0474 Good SG A NARROW 5 RANGE LEVEL LT0475 Good SG A NARROW 6 RANGE LEVEL LT0476 Good PRESSURIZER 7 LEVEL LT0459 Good PRESSURIZER 8 LEVEL LT0460 Bad Time PRESSURIZER 9 LEVEL LT0461 Good Data Historian September 2021 www.ams-corp.com SLIDE 105 OF 14

OLM Checks Calibration Over Much of a Transmitter Operating Range CRS039A-02 90 80 70 60 Pressure 50 Startup 40 Data 30 20 10 0

Reading 0 1000 2000 3000 4000 CRS039A-01 80 70 60 50 Pressure 40 Shutdown Data Time 30 20 10 0

0 200 400 600 800 1000 1200 1400 Time (Minutes)

September 2021 www.ams-corp.com SLIDE 106 OF 14

OLM Analysis at Startup September 2021 www.ams-corp.com SLIDE 107 OF 14

Commercial OLM Implementations in Nuclear Facilities Sizewell B : 2005 - present Sizewell B

Approved by UKs Nuclear Installations Inspectorate - 2005

Calibration induced human error Vogtle problems minimized Vogtle Units 1 and 2 : 2018 - present Advanced Test Reactor Advanced Test Reactor : 2015 - present September 2021 www.ams-corp.com SLIDE 108 OF 14

AMS OLM Topical Report

Step-by-Step OLM Implementation Methodology

- Determine Calibration Intervals

- Establish OLM Limits

- Perform Drift Analysis

- Perform Dynamic Failure Mode Assessment

Example Changes to Existing Tech. Specs.

- Description of OLM Program

- Changes to SR Frequency Column

- Changes to Bases Section of Each SR September 2021 www.ams-corp.com SLIDE 109 OF 14

Example Calibration Schedules for a Group of 4 Redundant Transmitters Cycle 1 2 3 4 5 6 7 8 9 10 11 12 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 Existing 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 1 1 1 2 2 2 Sizewell 3 3 3 4 4 4 1

AMS OLM TR September 2021 www.ams-corp.com SLIDE 110 OF 14

History of Transmitter Calibration Interval Extension EPRI ICMP Millstone (1989) 1990 NUREG/CR-6343 McGuire 1995 EPRI TR-104965-R1 NRC SER 2000 Sizewell B Regulatory Approval 2005 2010 2015 PWROG PA-SEE-0625 (2019) 2020 AMS-TR-0720 (2021)

September 2021 www.ams-corp.com SLIDE 111 OF 14

Thank You!

Questions?

Use Cases of Digital Twin Enabling Technologies in Nuclear Power Plants Technical Session - September 15, 2021 Moderator: Gene Carpenter, DOE Matthew Yarlett, Westinghouse Koushik A Manjunatha, INL Christophe Duquennoy, EDF Richard McGrath, EPRI Iikka Virkkunen, AALTO

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Matthew Yarlett - PKMJ Technical Services, LLC 2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 15, 2021 2

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Project Objective Integrate advancements in online monitoring and data analytic techniques with advanced risk assessment methodologies In November 2018, the U.S Department of Energy selected PKMJ Technical Services, Idaho National Laboratory, and Goals Public Services Enterprise &

Group (PSEG) Nuclear, LLC

1. Risk-informed approach to optimize equipment maintenance for an Advanced Nuclear frequency Technology Project.

2. Risk-informed approach to condition-based maintenance https://www.energy.gov/ne/articles/us-advanced-nuclear-technology-projects- 3. Develop and demonstrate a digital, automated platform receive-18-million-us-department-energy Duration August 2019 - July 2021 3

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Scope 4

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Potential Benefits 5

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Target Equipment Selected 60 Vibration Sensor Nodes have been installed across Salems 12 Circulating Water System pumps, motors and associated bypass valves. Each sensor node consists of two accelerometers sensitive to orthogonal in-plane motion and a temperature sensor.

6

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Wireless Vibration Sensor Installation 7

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Preventive Maintenance Optimization Recommended Equipment Task Title Current Frequency Industry Average Recommendation Frequency Refurbishment 6 years 14 years Less Frequent 9 years Pump External Visual Inspection 18/24 months 2.8 years Keep 18 months Vibration Analysis 3 months 5.5 months Less Frequent 6 months Oil Analysis 6 months 8 months Keep 6 months Motor Inspect/Electrical Testing 3 years 3 years Keep 3 years Replace Motor 6 years 10.7 years Less Frequent 9 years Motor Cable VLF TAN-Delta Testing 6 years 7 years Keep 6 years Protective Relays Inspect/Calibrate 6 years 4 years Keep 6 years Pressure Switch Calibration 4 years 4.2 years Keep 4 years

Risk models developed by INL support the pump refurbishment and motor replacement frequency recommendations

$4.37M - Net savings over next six years if implemented at the site 8

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Work Order Data Analysis

WO Failure Classification was performed using Natural Language Processing (NLP) techniques to classify work orders associated with equipment degradation, failure, and other. NLP was also utilized to evaluate the primary object types (component types), conditions identified, and actions performed for work orders at Salem (see bottom figure).

Created a plant process data and work order dashboard in PowerBI (see top figure). This was used for engineering review of events and work orders to better understand the impacts to plant process parameters when faults occurred.

Developed WO work type classifier to determine when work requires part usage, improves health, includes inspection (check-up), identifies potential issues, restores condition to As Good As New, or others.

Developed parts clustering technique to identify common parts used for work orders of a given type. This method allows for the identification of required and contingent parts, which can be used to enhance supply chain decision making based on future planned work.

9

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Identification of Fault Signatures

PSEG Plant Process data was used to identify characteristics of specific faults that occurred with the Salem Circulating Water System equipment

Time-domain and frequency-domain features were examined where possible to associate plant conditions with engineering characteristics

Fault Signatures were used as the foundation for developing diagnostic models

The development of fault signatures considered various scenarios based on the amount of indication available (i.e.,

without motor current, without vibration data, etc.) to provide insight into the flexibility of a fault signature solution 10

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Accomplishments - Development of an Automated Digital Platform

PKMJ developed a cloud-hosted application using Microsoft Azure to enhance industry decision-making and data visualization

Tools and services such as Azure Active Directory, Azure API Manager, Azure Function/Logic Apps, Azure DataLake Storage, DataFactory etc. were used to provide the structure, security, and data for the application

The digital platform developed under the project supports maintenance strategy optimization, responding to fault signature alarms, and generation of automated work orders.

11

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Summary 12

Integrated Risk-informed Condition-based Maintenance Capability and Automated Platform Future Development

PKMJ is in discussion with INL to continue the research, development, and commercialization of the digital platform solution. Areas of interest include:

Enhancements to Fault Signature Models

Integration of the Platform to Utility Systems

Improved Economic Modeling

PKMJ is also working with its parent company, Westinghouse, to incorporate other Condition Monitoring solutions into the digital platform 13

QUESTIONS?

Contact Information Matthew Yarlett - PKMJ Technical Services Email: Matthew.Yarlett@westinghouse.com Vivek Agarwal, PhD - Idaho National Laboratory Email: Vivek.Agarwal@inl.gov Harry Palas - Public Service Enterprise & Group (PSEG) Nuclear Email: Harry.Palas@pseg.com 14

References

K. Technologies, "KCF Vibration Sensor Node SD-VSN-3," [Online]. Available: https://www.kcftech.com/smartdiagnostics/products/sensors.html. [Accessed May 2020].

KCF Wireless Interference and SmartDiagnostics," [Online]. Available: https://www.kcftech.com/smartdiagnostics/resources/whitepapers/Application%20Note%20-

%20Wireless%20Interference%20and%20SmartDiagnostics.pdf. [Accessed May 2020].

E. Seneta, Non-negative matrices and Markov chains, 2nd ed., vol. XVI, (Originally published by Allen & Unwin Ltd., London 1973), 1981, p. 288.

K. S. Trivedi, Probability and Statistics with Reliability, Queueing, and Computer Science Applications, New York: John Wiley & Sons, Inc., 2002.

G. Bolch, S. Greiner, H. de Meer and K. S. Trivedi, Queueing Networks and Markov Chains, 2nd ed., John Wiley, 2006.

R. Nelson, Probability, Stochastic Processes, and Queueing Theory, Springer-Verlag, 1995.

Serverless Web Application, Published by Microsoft [Online], https://docs.microsoft.com/en-us/azure/architecture/reference-architectures/serverless/web-app. [Accessed July 2021]

15

September 15, 2021 Koushik A. Manjunatha Staff Research Scientist, INL Artificial Intelligence (AI) Driven Scalable Condition based Predictive Maintenance Strategy

Vision of Risk-informed Predictive Maintenance (PdM)

Strategy 17 V., A Manjunatha, K et al. (2019, September). Deployable Predictive Maintenance Strategy based on Models Developed to Monitor Circulating Water System at the Salem

Agarwal, Nuclear Power Plant (INL/LTD-19-55637). Idaho National Laboratory.

Data Analytics at Scale and Digital Twin Condition based PdM Flowchart A schematic representation of a Digital Twin module 18 V., A Manjunatha, K., et al. (2021, August). Scalable Technologies to Achieve Condition-Based Predictive Maintenance Enhancing the Economic Performance of Operating Nuclear

Agarwal, Power Plants (INL/EXT-21-01418). Idaho National Laboratory.

Physics based Modeling Numerical model representation of the top and bottom bearing housings, where mechanical loads are introduced Simplified numerical model for motor, pump, and shaft coupling Physics Driven model:

Requires: 1. domain knowledge (material, size etc.)

2. close interaction with operators 19

Condition based PdM Model and Outputs Vibration Data A representative figure of a particular fault captured in vibration signal Beginning of Fault Heathy state A schematic representation of a motor and pump with temperature measurement locations Fault progression period Forecasting dominant fault signature after prediction of a fault

Agarwal, 20 V., A Manjunatha, K., et al. (2021, August). Scalable Technologies to Achieve Condition-Based Predictive Maintenance Enhancing the Economic Performance of Operating Nuclear Power Plants (INL/EXT-21-01418). Idaho National Laboratory.

Scalable Condition based PdM Scalability is defined as expanding capabilities of a target entity to meet current and future application-specific requirements For NPPs, building a comprehensive AI model is challenging because

Faults are rare events, and it is highly unlikely for all the faults to occur in each component;

For a newly installed component/system or plant unit, Risk-Informed it is infeasible to build AI models from scratch; Predictive Maintenance Scalability

Collecting data at a centralized location is limited by Framework High bandwidth costs Real-time decision Privacy, security, and commercial concerns Agarwal, V., A Manjunatha, K., et al. (2021, August). Scalable Technologies to Achieve Condition-Based Predictive Maintenance Enhancing the Economic Performance of Operating Nuclear Power Plants (INL/EXT-21-01418). Idaho National Laboratory.

Federated Transfer Learning

Individual component-level model using component-specific available data sources

Consolidating the knowledge gained from individual component models into a master model,

Using the master model to make diagnostic and prognostic estimations of the entire system,

Applying (i.e., transferring) the master model to similar plant systems, either at the same plant site or at different plants.

Agarwal, V., A Manjunatha, K., et al. (2021, August). Scalable Technologies to Achieve Condition-Based Predictive Maintenance Enhancing the Economic Performance of Operating Nuclear Power Plants (INL/EXT-21-01418). Idaho National Laboratory.

Few Challenges to Consider Team

INL Team

Public Service Enterprise Group, Nuclear LLC Team

PKMJ Technical Services Team

Contact Information:

Koushik A. Manjunatha koushik.manjunatha@inl.gov Idaho National Laboratory 24

THERMAL PERFORMANCE MANAGEMENT FOR NUCLEAR POWER PLANTS WITH DIGITAL TWINS Use Cases based on EDF experience 09/15/21 Dr Christophe Duquennoy Nuclear Fleet Thermal Performance Expert christophe.duquennoy@edf.fr

THERMAL PERFORMANCE MANAGEMENT FOR NUCLEAR POWER PLANTS WITH DIGITAL TWINS Use Cases based on EDF experience 09/15/21

AGENDA

1. INTRODUCTION

2. DIGITAL TWINS FOR THERMAL PERFORMANCE

3. METROSCOPE SOFTWARE

4. USE CASES Titre de la présentation l jj/mm/aaaa l 27

DTG, from sensor to added value Asset Management and O&M support through expertise and advanced DATA acquisition diagnosis and prediction Dr Christophe Duquennoy Nuclear Fleet Thermal Performance Expert christophe.duquennoy@edf.fr

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE Thermodynamic Performance Testing and monitoring

- Thermodynamic measurement for NPP

- Performance monitoring, production and maintenance optimization l 29

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 30

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 31

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 32

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 33

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 34

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 35

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE l 36

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE Saint Laurent NPP: Bypass leak on cooling tower vs Row Water leak in condenser We provided the maximum affordable power to allow to repair Row Water leak within operation. Avoided Plant Shut down Cruas : Diagnostic for 14 Mwe lost in R501 HP Heater Medialization of simultaneous defects : Low level in heater + condensates leak + vapor entrainment l 37

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE Nogent NPP draining pump rupture:

Weve got a drain piping under our water collectors which is about to break. Our analysis lead to maintenance and Should we repair it before we turn on the avoided ~600 k of loss.

plant ?

Realistic model of cooling tower l 38

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE AI l 39

INTRODUCTION DIGITAL TWINS FOR THERMAL PERFORMANCE AI l 40

METROSCOPE SOFTWARE - GENERAL PRINCIPLE METROSCOPE SOFTWARE AI Software Universal Block The software combines a Digital Twin of the process and AI to perform a diagnosis in operations

+ =

The software uses existing information from sensors to analyze the plant performance and diagnose underlying causes impairing the Digital twin Specific block expected behavior of the process l 41

METROSCOPE SOFTWARE - GENERAL PRINCIPLE Three steps to perform a Diagnosis 1 Metrological analysis of the raw data 2 The symptoms are generated using the Digital Twin.

3 The diagnosis is produced thanks to AI Two bricks to make a Digital Twin

the nominal model represents the nominal behavior in operation of the process, calibrated on historical data

the failure library is a mathematical description of the failure modes of the process and their impact on the measurements l 42

METROSCOPE SOFTWARE - AI METROSCOPE AI belong to the domain of Symbolic AI (in opposition to Statistical AI)

It is meant to address Small Data problematics where decision relies on both Knowledge and Data.

METROSCOPE AI has been inspired by Medical Diagnosis Most knowledge-based systems have two distinguishing features: a knowledge base (here the digital twin) and an inference engine (here the Bayesian network)

Awarded Best Innovation in Nuclear in 2019 by the Société Française de lEnérgie Nucléaire Illustration des chanes de Markov l 43

METROSCOPE SOFTWARE DIGITAL TWIN The Digital Twin (DT) is a numerical representation of the secondary side, able to simulate both nominal and impaired behaviors of the process.

DT is built from the P&ID of the plant and calibrated on historical data.

DT of a NPP is composed of around 10 000 equations and variable declarations.

Perimeter and Accuracy of the DT are meant to evolve overtime Jumeau numérique du circuit secondaire l 44

METROSCOPE - DEPLOYEMENT ON EDF FLEET Fleet

References:

58 condensers digital twins encapsulating the best ever seen performance based on historian data.

CP2 32 cooling towers digital twins tuned CP1 encapsulating the best ever seen performance based on historian data Métroscope compatible Digital Twins Production of all the reference models for French Nuclear Fleets METROSCOPE : 56 units in 3 years Production of all defect libraries based on EDF operation feedback.

H4 - P4 N4 In particular: Expertise on key components in terms of thermodynamic performance:

Condensers, Cooling Towers, Heaters.

l 45

HIGH PRESSURE HEATER LEAK Detection of simultaneous leaks on both MSR and HP Heaters : Performance Gain

MSR condensate regulation fault was detected by the operator (since 2015 November 30th)

Avoidable loses 0,5 GWh

Meanwhile heaters leak on condensates occured (2015 December 14th). It was not detected by operators.

HP Heater 602 leak

A lost of 2 MW between December 15th and 27th would have been avoided with METROSCOPE (not deployed at MSR tank 101 leak this time)

CONDENSATE COLLECTOR TANK LEAK

Early detection of small magnitude leak (~2% of the nominal flow) Performance

Reliable quantification of the impact (1.5 MWe)

Diagnostic of multiple simultaneous defects Avoidable loses ~ 8 GWh First detection by Partial METROSCOPE Leak of Leak Repaired Detection by operators resolution ACO0027VL valve (old fashion method.)

METROSCOPE provision Full resolution 01/31/18 09/04/19

CONDENSER LOSES

Simultaneous Multiple Diagnostic

Accurate Quantification of condenser losses due to Performance updated reference Digital Twins Avoidable losses 1 mbar = 1MWe

CONDENSER LOSES

Real time reliable monitoring of condenser thermal performances Performance / Maintenance

Possibility to optimize biocidal injection during crises

Accurate Quantification of condenser losses due to updated Avoidable losses 1 mbar = 1MWe reference Digital Twins Optimization of operating solutions :

Biocidal injection, Taproges CTA, reboiling.

HEATERS TUBE RUPTURE

Early detection (2 or 3 weeks) Maintenance

Real time monitoring of number broken tubes

Maintenance schedule optimization

Limit the duration of maintenance intervention (better prepared) Gain Optimization of

Capitalization of feedback of evolution signatures. maintenance scheduling and duration First Tube rupture Tubes plugging

LOW HP HEATERS LEVEL

Situation before METROSCOPE not seen and solved by operators Maintenance

Detection of low level situations by METROSCOPE Gain Plant adjustment Low Heater Level detection

HP HEATERS LEVEL REGULATION

Real time detection of low level situations by METROSCOPE

Optimization plant adjustment by operators Maintenance Gain Plant adjustment Disturbed Heaters level Adjustment of HP Heaters regulation niveau level regulation

MSR VENTILATION OBSTRUCTION

Based on low signal detection (variation under 4% = 2 kg/s)

After operator intervention, the defect is solved Maintenance Gain: Incondensable gas management MSR Ventilation abnormally closed (GSS009VV) MSR Ventilation re-openned METROSCOPE commissionning 06/21/19

MSR VENTILATION LEAK

Simultaneous Multiple Diagnostic Performance

Almost MWe Losses are explained (~ 2 MW of residual losses)

Avoidable loses 0.4 GWh

CONCLUSION SYNTHETIS OF MAIN GAINS A large variety of situations are well catch

- Condenser Performance drift

- Steam flow decrease in the turbine

- Feed Water Heater unoptimal level regulation

- Feed water tank leak METROSCOPE automated diagnosis strongly simplifies operators life METROSCOPE is online, and provide a fleet wide supervision of the plants (enhance the production national reports for LTO)

Real life situations leads to

- Validation of DT (Reliable reference of best historical performance)

- Validation of Failure Libraries (Know ledge capitalization)

CONCLUSION SYNTHETIS OF MAIN GAINS Performance Monitoring

- Identification of major failures in terms of losses

- Smaller MW losses are catch

- Some failures impact maintenance only With METROSCOPE unexplained MW losses are

- Under 2 MW 53% of the time

- Under 3 Mw 70% of the time Here METROSCOPE has been calibrated to detect just a little bit more than whats really happening METROSCOPE interface is very useful to analyze unexpected behaviors of the tool

CONCLUSION EXTENSION OF AUTOMATIQUE DIAGNOSTIC l 57

CONCLUSION EXTENSION OF AUTOMATIQUE DIAGNOSTIC l 58

CONCLUSION TOMORROW - COOLING TOWER MONITORING WITH METROSCOPE Numerical model plugged in the Metroscope for live diagnosis AIl diagnosis from physical Simulation l 59 www.metroscope.tech

CONCLUSION TOMORROW - COOLING TOWER MONITORING WITH METROSCOPE Develop Defect models Use available data to validate l 60 www.metroscope.tech

CONCLUSION TOMORROW - COOLING TOWER MONITORING WITH METROSCOPE SYNTHETIC OVERVIEW OF SERVICES PROVIDED BY SPA ESTIMATION OF THE SERVICE PROVIDED LEVEL OF INVOLVEMENT GAIN 5 GWh/ major defect trained operator + time 20 GWh/other defect required for level 1 support 4 major defects/year + functional SPA analysis 20 GWh/maintenance competent support engineers

+ time required for level 2 3 situations /years support competent engineering unit+

Difficult to quantify Time required to go further in the analysis Gain of 0,2°C 160 competent engineering unit+

GWh over 30 years + expertise on operation+

design engineering l 61

CONCLUSION TOMORROW - COOLING TOWER MONITORING WITH METROSCOPE SYNTHETIC OVERVIEW OF SERVICES PROVIDED BY SPA + METROSCOPE ESTIMATION OF THE SERVICE PROVIDED LEVEL OF INVOLVEMENT GAIN 5 GWh/ major defect trained operator + time 20 GWh/other defect required for level 1 support 4 major defects/year + functional SPA analysis 20 GWh/maintenance competent support engineers

+ time required for level 2 3 situations /years support competent engineering unit+

Difficult to quantify Time required to go further in the analysis Gain of 0,2°C 160 competent engineering unit+

GWh over 30 years + expertise on operation+

design engineering l 62

THANK YOU!

Any questions?

l 63

APPENDIX l 64

DTG, from sensor to added value Asset Management and O&M support through expertise and advanced DATA acquisition diagnosis and prediction Dr Yves-Laurent BECK Fleetwide eMonitoring Program Manager Dr Christophe Duquennoy Global Product Manager on Power Block Monitoring Nuclear Fleet Thermal Performance Expert yves-laurent.beck@edf.fr christophe.duquennoy@edf.fr

EDF-DTG Our mission FROM DIAGNOSIS TO FORECASTS 3 MAIN FIELDS :

MEASUREMENT INSPECTION EXPERTISE

> Measurement > Monitoring on behalf > Diagnosis, pronostics engineering, metrology, of operators & consulting for data bases operators and maintenance teams

Monitoring, diagnosis and forecasts for EDF Group power plants (nuclear, hydro, fossil-fired and wind energy) to assist operators in making effective safety decisions and in managing performance

Develop technical skills and expertise required to operate energy production facilities, to find solutions about energy development and environmental issues

Integrate and make reliable new R&D monitoring solutions

EDF-DTG Our customers Power generation Hydro Nuclear Institutionals Renewables Industry Thermal

EDF-DTG Our strength Accurate and independant diagnosis & prognosis DATA historian based on 50 years of experience Know-how to convert DATA into value An unique expertise in O&M Innovative solutions

DESIGN & O&M Support Services MECHANICAL INSPECTION AND SYSTEMS AND VIBRATING ELECTRICAL EQUIPMENT DIAGNOSIS ENGINEERING

Inspection during production

Monitoring mechanical

Diagnosis of main electrical behaviour of power plants, equipment needed for

Assessment of key static pipe systems and rotating power plants operation, components for power machinery plants (penstock, secondary based on tests carried out circuit): non destructive

Acoustic testing of primary on site and in the lab testing, corrosion-erosion circuit for nuclear fleet modeling, damage analysis

Diagnosis using infrared and repair tracking

Condition-based thermography (electrical maintenance of nuclear trouble spots) valves 69

DESIGN & O&M Support Services SETTINGS, PROTECTION, ACOUSTICS THERMODYNAMIC DIAGNOSIS OPTIMISATION PROCESSES,

Noise control for all EDF AND PROGNOSIS ANCILLARY SERVICES Group facilities : acoustic

Testing the safety protection impact studies and

Measuring thermodynamic of power-generating modeling, sound power performances of thermal facilities measurements in situ, power plants (fossil-fired and nuclear)

Testing the network

Optimising the reconstitution and voltage recovery following an soundproofing of

Monitoring energy incident installations and to ensure performances, looking for compliance with regulations productivity gains

System services monitoring and optimisation:

contribution to system voltage and frequency 70 stability

DESIGN & O&M Support Services METROLOGY, IT & DATA MONITORING NUCLEAR WATER RESOURCE FORECAST technologies FACILITIES & MANAGEMENT

Development of appropriate

Continuous monitoring of IT solutions for the various mechanical behaviour of

Evaluation and management business activities equipment and civil of the impacts of electricity engineering structures generating activities on

Data collection, atmospheric, acoustic and management, processing

Containment tightness : aquatic environments and archiving monitoring and control during pressure tests

Impact of aquatic

Certified by Cofrac, the DTG environment on electricity laboratory is a reference for generation the calibration of temperature, pressure and 71 humidity variables

Digital Twin of a Real-time Radiation Monitoring Network An EPRI NextGen RP Project Rich McGrath Principal Technical Leader rmcgrath@epri.com 2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 14-16, 2021 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Discussion Topics Background of Need for Digital Twins Results of Scoping Study of Available Technologies In Progress Project Tasks Future Work 73 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Background of Need for Digital Twins Utility RP Organizations need to seek additional measures for cost reduction while still maintaining excellent performance and safety while faced with the following challenges:

- Shrinking contract RP technician resources,

- Reductions in site RP staffing,

- Cost efficiency,

- Knowledge retention and transfer, and

- Maintaining and/or enhancing worker safety.

Currently, most radiation protection (RP) tasks for the collection of radiation field data are conducted manually at nuclear power plants. For example:

- Most radiation field measurement and characterization activities (i.e., surveys) are conducted

- Data from radiation surveys are manually entered into the plant database and subsequent data analyses to inform radiation protection, and/or source term management, and/or ALARA planning are conducted manually.

The EPRI Digital Twin Project is evaluating:

- Technologies that provide remote/automated radiological measurements that accurately reflect the radiological environment in the plant

- If technologies are available, then creation of a Digital Twin of the radiological environ can be used to optimize work in the radiological areas of the plant.

Digital Twin of a Real-time Radiation Monitoring Network Phase 1 - Demonstrate a Geophysical Application for Analyzing Radiological Example of Actual Interpolation of Survey Data Survey Data Increased use of remote radiation monitoring technology is providing real-time radiological data.

Measurement points are shown as single point sources with no values in between measurements.

Geospatial technologies are available that can interpolate available remote detector readings to model the radiological conditions between measurement points.

Applications could be applied in reverse to determine where remote monitoring devices should be placed to adequately monitor dose rates This EPRI project will in 2021:

- Evaluate available software applications

- Test geostatistical software tool(s) to see if tool accurately interpolates actual survey data Survey Measurements Interpolation of Survey Data 75 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Initial Tasks of EPRI Project EPRI Team has completed an initial scoping study to determine what software tools and associated technologies are currently available or planned that can create a digital twin of the radiation fields at a nuclear power plant.

For each candidate software/technology supplier, the technology search attempted to determine the following:

- The current state of development and/or planned future enhancements as it relates to this project.

- Current state of deployment - where it has been used or is being used at a nuclear power plant site

Examples of Supplier Capabilities for Tabletop Demos EPRI Indoor Positioning System (IPS) system demonstrations showed functionality applicable to digital twin technology such as:

- Track tags to ~1-2 meter accuracy

- Collect dose rate data continuously

- Generate live dose rate maps EPRI Demonstrations Performed at Mirion Orion RTLS nuclear power plants:

- Bluetooth Low Energy (BLE) based Quuppa

- Ultra Wide Band (UWB) based Mirion Orion Real Time Locating System (RTLS)

Quuppa Demonstration Workers provided with positioning tags and teledosimeter Anchors with magnet mounts for receiving signals from dosimeters installed as shown in images Dose rate data from teledosimeter and positioning data from tags are merged together to create dose maps 78 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

MIRION ORION' RTLS DEMONSTRATION Anchor Placement

Similar approach but different anchor design for Mirion system 80 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

MIRION ORION' RTLS DEMONSTRATION Accuracy

Accuracy was observed to be 1-2 m

For some isolated locations, the position was erratic due to structural interference

Accuracy increases with number of anchors and transmit frequency

Dose rate data displayed on the maps 81 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Other Technology Available Capabilities offered by other vendors:

- Creation of 3D maps of the plant using LIDAR system incorporated into the portable radiation detector Readings from detector automatically recorded onto the 3D map as the survey is conducted Upload new or existing 3D CAD generated maps of the plant into the software:

- Radiation readings automatically loaded onto maps from:

EPRI Digital Twin Project Status - Phase 1 Multiple vendors have:

- At least one technology that appears to be capable of supporting development of a digital twin of a radiation monitoring network, or have much of that development in place.

- Appear to have products that are fairly mature with variations on aspects of their product relative to each other (e.g., location tracking, radiation field heat map, use of fixed radiation monitors, etc.).

- Have products that were able to utilize 3D scanning technology that would subsequently support area mapping, heat map generation and location tracking.

- Expressed a willingness to participate in a tabletop demonstration Selection of supplier(s) for the Tabletop Demo this month Demos to be conducted in 2021 83 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Digital Twin of a Real-time Radiation Monitoring Network Phase 2 - Apply Machine Learning to Geophysical Radiological Survey Application Machine learning can be coupled with the geospatial algorithm to refine the radiation field estimates as measurements are updated The digital twin could be used to provide ongoing monitoring, trending, and alerts and allow for alternative maintenance and dose optimization scenarios to be investigated in cyberspace before the work is performed Simulation results could be visualized by the worker during job preparations, job briefings, and in the work environment using augmented reality techniques.

Simulations would enable development of efficient maintenance practices to save time, reduce worker exposure, and reduce cost Could be used for event recreation and emergency plan exercises The proposed added scope for 2022-2023 includes:

- Evaluate software tool(s) for use in machine learning of survey data Example of ALARA Planning Tool for

- Test application to see if tool accurately predicts future radiation dose rate trends from survey data analyzed Potential Further Development 84 www.epri.com © 2021 Electric Power Research Institute, Inc. All rights reserved.

Iikka Virkkunen Prof. (Adjunct) at Aalto University, Managing Director or Trueflaw, Ltd.

NDE4.0 and Machine Learning for In-service Inspections

NDE increases reliability Todays NDE3.0

Qualified and reliable

Rich digital data

Important contribution to reliable and safe operation But

Time consuming

Potential for human errors

Limited information extracted

More is less

Machine learning enables human-level automated evaluation Human-level performance in automated defect recognition Applicable to various inspections:

Ultrasonics Digital radiography Already in field use in non-nuclear inspection

Case: GKN digital X-ray of aerospace welds High quality automated welds Small flaws, difficult to detect Current status:

Human level detection

Accurate sizing

Criteria comparison Additional data on small flaws

Case: CRDM TOFD inspection EPRI - Trueflaw collaboration Very difficult data Limited real data available Limited real flaws available Human level performance Next: field test planned

NDE3.0 + ML = NDE3.5

Automated analysis can manage large data

Inspector work elevated to focus on the important parts

More is more again Additional benefits

More sensitivity

Predictive capability

Added value from NDE

NDE4.0 Integrated NDE data

Connected systems can aggregate data

Digital twin etc.

Elevated to focus on the important parts for the system

More is exponentially more due to network effects Additional benefits

Situation awareness

Predictive capability

Added value from NDE

How do we get there, safely?

Compatibility with existing qualification Compatibility with current practices Data security

Qualification 2020 ENIQ published a position paper:

Qualifying (certain) ML systems filled within framework ENIQ RP13:

Recommended practice for Qualification of Non-Destructive Testing Systems that Make Use of Machine Learning

We now know how to qualify ML Repeatability of ML systems is good for qualification Main changes are with test data

ML systems may require more test data

Combining open and blind trials may reduce test block needs Qualifications must be done using frozen software

Edge devices can secure data

Stand-alone unit

Easy to use

Works with existing data files

Multiple options for reports

Can provide traditional reports

Can integrate to digital twins

Conclusions For digital twins, NDE 4.0 The tools are ready:

Machine learning

Edge computing The transition is clear

Qualification RP13

Adoption with stand-alone units

Digital Twin Enabling Technologies in Advanced Reactor Applications Technical Session - September 15, 2021 Moderator: Angela Buford, NRC Ian Davis, X-energy Nurali Virani, GE Anthonie Cilliers, Kairos Power Emilio Baglietto, MIT Michael Buric, DOE Bob Urberger & Roger Chin, Radiant

Digital Twin Enabling Technologies in Advanced Reactor Applications Xe-100 Digital Technologies Overview Ian Davis, Senior Digital Twin Systems Engineer September 15, 2021

UCO TRISO Particle - Primary Fission Product Barrier 0.855 mm 19 000 0.425 mm TRISO coated 60 220 000 UCO kernel particles in a mm pebbles in the core Porous Carbon pebble Pyrolytic Carbon Silicon Carbide Pyrolytic Carbon Primary safety goal is to ensure that fission products are retained within the TRISO coated fuel particles to the maximum extent possible This is achieved through production of high quality TRISO fuel and ensuring that temperatures in the core never exceed the temperatures for which the fuel has been tested (AGR Experiments)

Background:

Xe-100 Plant Overview Standard X-energy plant have 4 Reactors

- 4 Turbines producing 320 MWe, attributes include:

200MWth/80MWe Per Module Process heat applications Proven intrinsically safe Meltdown proof Walk-away safe Modular construction Requires less time to construct (2.5-4 years)

Road transportable for diverse geographic areas Uses factory-produced components Load-following to 40% power within 15 minutes Continuous fueling; resilient on-site fuel storage

Xe-100 Digital Twin Tools 3D Models with AR / VR Operator Training Simulator Plant Historian AI / ML Models

ARPA-E GEMINA Project Progress Summary

Project

Title:

Advanced Operation & Maintenance Techniques Implemented in the Xe-100 Plant Digital Twin to Reduce Fixed O&M Cost

$7.5 Million award from DOE for Digital Twin (DT) and Central Maintenance Model (CMM) concepts

Simulator and 3D Models

Systems Engineering Transition Concept Requirements System System Design System Test &

Operation &

Development Engineering Architecture & Development Integration Evaluation Maintenance

Systems Engineering with Xe-100 Digital Twin (Now) Design Transition Concept Requirements System System Design System Test &

Operation &

Development Engineering Architecture & Development Integration Evaluation Maintenance Operations

& Maintenance

Systems Engineering with Digital Twin: Maintenance Transition Concept Requirements System System Design System Test &

Operation &

Development Engineering Architecture & Development Integration Evaluation Maintenance Ex: What is Maintenance maintenance burden Analysis for Feedwater pumps?

Ex: Analysis suggests additional Feedwater train be added for redundancy.

Analysis suggests additional instruments be added for better monitoring coverage.

Systems Engineering with Digital Twin: Security Xe-100 Design Docs Transition Concept Requirements System System Design System Test &

Operation &

Development Engineering Architecture & Development Integration Evaluation Maintenance Ex: What does an Security attempted sabotage Analysis event look like?

Anomaly Detection with Machine Learning Deep Neural Networks (DNNs) to support plant operation Long Short Term Memory Autoencoder (LSTM-AE)

Q: How soon will any setpoint be Q: Is the system on normal exceeded?

operation (steady-state)? Goal: predict time-to-threshold Goal: detect anomalies with minimal delays Setpoints: helium T & P, power, steam T & P, mass flow rate Long Short Term Memory Dense (LSTM-D) (not shown)

Transient Corrective events Transient Progression operations Anomaly detection characterization prediction Human Machine Interface (HMI)

Q: What is the ongoing transient?

Goal: categorize the transient Convolutional Neural Network (Convnet)

Major events

Circulator pump trip

Condenser pump trip

Control valve position

Clogging of MSV in turbine Time

Control rod withdrawal

Feedwater pump trip

Unknown

2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization Wednesday, September 15th Digital Twin Enabling Technologies in Advanced Reactor Applications Humble AI for Reliable Machine Learning-based Health Twins Dr. Nurali Virani Lead Scientist - Machine Learning, GE Research December 7, 2021

What are Machine Learning-based Health Twins?

Digital twins for fault detection, diagnosis, and health estimation of physical systems and components 1

Detection x

Fault or anomaly detection with early warning is formulated as an Unsupervised classification unsupervised learning problem

Needs data of nominal behavior from modeled system 2

1 Diagnosis

Fault classification is formulated as a supervised learning problem

Needs labeled data for multiple fault classes Supervised classification 2

Health estimation & Forecasting

Health estimation and performance prediction of continuous-valued variable is formulated as supervised learning regression problem Regression

Needs input-output data pairs for training the model Key gap in using health twins-based automation for critical industrial infrastructure is in characterization of reliability and trust as safety and performance are paramount 111

Understanding prediction reliability can help improve outcomes and build trust in AI Automated Fault Classification Need Human Help The FMCRD behavior is anomalous. I dont know if I know this is fault class 1, here is my FMRCD has a known fault, evidence, please authorize to take partner can you appropriate actions. investigate it and trigger The feedwater pump might have fault 1 safe mode operation.

or 3, but not 2, send technicians who can address type 1 and 3 faults Image source: FIG. 2.2 in https://aris.iaea.org/PDF/ABWR(Hitachi-GE)_2020.pdf Reduce O&M cost via automation, Better collaborative outcomes and ability minimize repeated crew trips and to deal with novel scenarios avoidable shutdown events

Surrogate model-based optimization and control needs better assessment of prediction reliability Balance of plant (BOP) Component Script: https://github.com/to-mi/gp-demo-js Fault Scenarios GP from live version at: http://www.tmpl.fi/gp/

Operational Potential Performance Risk Surrogate Assessment (OPRA) data model# 2 Demand power Reconfiguration optima Level changes Alternatives true function Backward Model 1 region of trust BOP settings region of reconfigurations Required Control Rod extrapolation density/positions 2 Reactor Physic-based Lack of coverage: Complete variability that we expect during Surrogate BOP deployment is not available in training data Model simulation BOP boundary conditions Can I trust the surrogate model prediction for the suggested From ARPA-E GEMINA Predictive Maintenance with Digital BOP boundary conditions and control rod density values?

Twin program (GE Research, ORNL, UTK, GE Hitachi, Exelon)

Humble AI for Digital Twin An AI that is aware of its own competence and improves its competence via learning Humble AI Prediction Outcome with Defining capabilities uncertainty Model Data ML understand region of trust competence And / or Model evaluation Prescription Control input or strategic action quantify uncertainty Robust baseline Continuous learning Safe exploration ask for help when incompetent Human Approval Run a simulation (optional)

Needs continual learning from 1 or more sources explainability and Ask a human interpretability Ask other agents Complex system Humble AI will reduce Time to Fall-back mode selection Value Humble AI will maintain safety

Key Gap and Research Question

Key Gap: Machine learning provides statistically impressive results which might be individually unreliable.

My validation accuracy was high, so trust my belief Soft-max value for predicted class is high, so trust my belief (distance from hyperplane)

In many situation, randomized inspection of samples is inadequate to verify reliable outcomes and complete inspection defeats the purpose of using AI Can we characterize individual prediction reliability to understand the limitations of ML due to:

Observability (or separability)?

Brittle extrapolation?

Support types to create justification for model competence Using model input, model internal representation, and model outputs for support (3) Geometric neighborhood in

Types of Support Guard against embeddings (latent spaces)

(4) Output uncertainty anomaly 1 Input anomaly

Input drift

Extrapolation 2 Reconstruction error

Input drift

Extrapolation 3 Geometric neighborhood in

Extrapolation embeddings

Ambiguity

Adversarial manipulations 4 Output uncertainty anomaly

Input drift

High process uncertainty 5 Residual error drift

Concept drift (1) Input likelihood with density (2) Reconstruction error-based estimation anomaly (5) Residual error drift encoder decoder Neighborhood operator: actual (a) by number of neighbors expected l l (b) by radius (distance threshold)

Model Input Model Internal Representation Model Output

Characterizing region of trust, overlap, and extrapolation for AI to get justification-based reliability Output = , , ,

Justification from supports

= ;

Justified belief is knowledge

= I know (Region of trust)

I may know (Region of overlap/ambiguity )

I dont know (Region of extrapolation)

1. Virani, N., Iyer, N. and Yang, Z. Justification-Based Reliability in Machine Learning. in AAAI 2020

2. Bhushan, C., Yang, Z., Virani, N. and Iyer, N., 2020. Variational encoder-based reliable classification. arXiv preprint arXiv:2002.08289. (IEEE ICIP 2020)

Examples of evaluating competence using internal representations Visualization of Latent Dataset: Space - Car and Truck CIFAR (car and truck class)

Classes - IK, IMK, and Belief : 9 Base model:

IDK IMK [4, 9] Residual Network Layer:

Truth: 4 Global average pooling Support:

-ball with 2 -metric Belief : 4 IMK [4, 9]

Truth: 4 Belief : 1 IMK [1, 7]

Truth: 7 Support from latent space as exemplars for IK [7] interpretability

Data to facilitate Health Twin development and validation Tennessee Eastman Process (TEP) simulation dataa Open-source benchmarking data of industrial chemical process for the purpose of developing, studying and evaluating process control and fault detection Process has 12 valves available for manipulation and 41 measurements available for monitoring or control.

Time series sensor data, with faults injected using simulation Variables sampled every 3 minutes for a total of 25 hours2.893519e-4 days 0.00694 hours 4.133598e-5 weeks 9.5125e-6 months (training data) and 48 hours5.555556e-4 days 0.0133 hours 7.936508e-5 weeks 1.8264e-5 months (test data)

Overall Data size: 15M rows x 52 variables x 1-label Labels: 20 fault-types + Normal a:

1> Downs, J.J. and Vogel, E.F., 1993. A plant-wide industrial process control problem. Computers & chemical engineering, 17(3), pp.245-255 2> Rieth, Cory A.; Amsel, Ben D.; Tran, Randy; Cook, Maia B., 2017, "Additional Tennessee Eastman Process Simulation Data for Anomaly Detection Evaluation" 119

Humble Health Twins for Fault Classification Approach & Outcomes 1: DATA PREPROCESSING 2: BASELINE HEALTH TWIN MODEL H4 44%

56%

Automated Seek help IK, IMK, IDK 3: HUMBLE HEALTH TWIN 4: OUTCOME

Summary Humble AI is an AI that is aware of its own competence and improves its competence via learning Current focus of Humble AI is on ML-based digital twins (specifically health twins)

Humble AI enables to characterize region of trust for health twins to get justification-based reliability where automation can be enabled Extend to use model input, model internal representation, and model outputs for support (beyond NN models)

GRCTHINK Thank You!

Building a world that works Collaborate with us on research programs for a better tomorrow (nurali.virani@ge.com) 122

ENABLING TECHNOLOGIES FOR DIGITAL TWIN APPLICATIONS FOR THE KP-FHR CHRIS PORESKY AND ANTHONIE CILLIERS 123

Kairos Powers mission is to enable the worlds transition to clean energy, with the ultimate goal of dramatically improving peoples quality of life while protecting the environment.

Confidential and Proprietary No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

Advanced reactors - challenges and opportunities

Industry experience in plant simulators is entirely focused on light water reactors

Challenging environments, more difficult component and structure qualification

Limited historical data on performance and documentation of material and chemical properties

Parallel engineering - developing capabilities while developing application space simultaneously

Advanced reactor priority phenomena may occur at different (i.e. slower) timescales

Modern instrumentation and control may more easily lend itself to data communication

Two-phase phenomena may be deprioritized for non-water coolants

Passive safety may reduce the dependency on active control Confidential and Proprietary 125 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

Overview of Kairos Power

KP-FHR Inherent Safety and Economic Potential are Unique:

Robust Inherent Safety Large fuel temperature margins fission products retained by fuel and primary coolant Low-pressure system Passive decay heat removal

Lower Capital Costs Reduced reliance on high-cost, nuclear-grade components and structures through FHR intrinsic safety and plant architecture KP-FHR HTGR Fast Breeder PWR SMR Reactor Leverage conventional materials, existing industrial equipment, Flibe Helium Sodium Water and conventional fabrication and construction methods 280 MWth 250 MWth 260 MWth 200 MWth 120 MWe 100 MWe 100 MWe 60 MWe

Improved Operating Economics

Reactor vessels drawn to notional scale.

High efficiency Flexible deployment of low-cost nuclear heat 126 Confidential and Proprietary No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

KP-FHR Safety Case - Defense in Depth Barriers

Radionuclide Retention Capabilities Confidential and Proprietary No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

KP-FHR I&C design Robust Plant protection and control Inherent Safety

KP-Shield: Passive, robust, reliable safety shut-down system

Low operating pressure Safety Hazard Intervention and Event Limiting Defense (Shield)

Large fuel temperature margins

KP-Sword: Active plant control

Effective passive decay heat removal System with Operational Reliability and Diagnostics (Sword)

Uniquely large heat capacity

KP-Heart: Intelligent Health Monitoring

Strong negative temperature Health Evaluation and Analysis in Real-Time (Heart) coefficient

KP-Sight: Semi-autonomous control room

Slow transient response Semi-autonomous Industrial Grade HMI Technology

KP-Bolt: Electrical supply Basic Ohm Law Triangle (V = I.R)

Confidential and Proprietary 128 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

Active plant Control & Health Monitoring: KP-Sword & KP-Heart Normal Operations and Anticipated Operational Occurrences KP-FHR Digital Twin Operating KP-FHR Plant experience Control sensor data stream Machine Selected Learning Corrective data action Update Plant Probabilistic Risk Assessment, Structures Systems &

Components performance (plant understanding) KP-Heart KP-Sword

~~~~~~~~~~~~~~~~~~~~

~~~~~~~~~~~~~~~~

~~~~~~~~

Control Optimize control Actuation actions Optimize control policy Verify Plant Health Emergency Operating Procedures, Severe Accident Management Guideline, Code of Federal Regulations

~~~~~~~~~~~~~~

~~~~~~

Confidential and Proprietary 129 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

KP-FHR digital twin working definition

Collects, compiles, and contextualizes plant information from a variety of sources and makes it accessible to aid plant operation Visualization Database Physics simulation Real-time communications and feedback Data analytics Plant health monitoring Operations optimization Operator decision support Confidential and Proprietary 130 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

KP-FHR digital twin development process

Small test facilities Demonstrate ability to connect facilities Prototypical using different hardware and software facilities Validate simulation tools against representative physical systems

Large test facilities Fundamental connectivity, basic use cases Deploy digital twins alongside tests to gather operating experience and train models Integrate simulation into operation to Small test improve understanding of system Large test facilities Real-time decision-making support and online plant facilities

Prototypical facilities health monitoring Support decision-making in real time Passively and continuously monitor plant health and flag unexpected behavior Operating experience and real-time integration of simulator tools Confidential and Proprietary 131 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

Desired capabilities and enabling technology

Desired capabilities Run continuously alongside plant operation Provide lookahead and optimization capabilities to operators and support staff Flag unexpected behavior and prompt important operations and maintenance activities

Enabling technology Communications infrastructure: talk to engineering tools, plant hardware, controls and simulation software Physical processes: develop and specify all plant components and prototypical information i: develop faster-than-real time physics simulation and data analysis capabilities Confidential and Proprietary 132 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

A note on cybersecurity and plant design

If the plant design precludes consequence, the risk for all failures is lower

If instrumentation and controls architecture precludes consequence, the reactor protection functions are not at risk

If reactor protection functions are not at risk, the value of added capabilities becomes much more attractive despite potential risks

Digital twins can bolster cybersecurity by improving the ability not only to detect but also to mitigate cyber threats Mitigation of cyber threats and approach to cybersecurity should be baked into plant design the same as it is for mitigation of insider threat, access control, intellectual property management, etc.

Confidential and Proprietary 133 No Reproduction or Distribution Without Express Written Permission of Kairos Power LLC

NSE Nuclear Science & Engineering at MIT science : systems : society HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS Focus on enabling technology Emilio Baglietto Massachusetts Institute of Technology Department of Nuclear Science and Engineering 134

NSE High Fidelity Digital Twins for BWRX-300 Nuclear Science and Engineering The Team Dr. Christer Dahlgren(co-PI)

Prof. Emilio Baglietto (PI) Prof. Koroush Shirvan (co-PI) Dr. Panos Tsilifis (co-PI)

Douglas McDonald David Hinds Yu-Jou Wang (PhD) Brandon Aranda (MS)

Genghis Khan (co-PI)

Charles Heck 135 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE The Backbone Nuclear Science and Engineering

MIT: e.g. demonstrated reduction of operating uncertainty through high fidelity simulations U.Otgonbaatar, E.Baglietto, Y.Caffari , N.E.Todreas and G.Lenci - A METHODOLOGY FOR CHARACTERIZING REPRESENTATIVENESS UNCERTAINTY IN PERFORMANCE INDICATOR MEASUREMENTS OF POWER GENERATING SYSTEMS - JVVUQ

GE: e.g. digital twin deployed in Nuclear and Aerospace Industries C

Digital Twin of BWR Stress Intensity Factor after simulated loss of feedwater pumps Digital Twin of Steam Dryer Stress Intensity Factors and Crack Lengths due to Vibration 136 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Why high-fidelity and why now Nuclear Science and Engineering Established Digital Twin

Advancement and ABWR demonstration of high-fidelity simulations based maintenance approaches and model based fault system detection techniques. Operating History Data Quantitative Data Assimilation Assessment for O&M

Address mechanical and BWRX300 thermal fatigue failure modes which drive O&M activities of BWRX-300

[NUREG/CP-0152]

High Fidelity Virtual Quantitative Operation Model Data Assimilation Assessment for O&M High Fidelity Digital Twin 137 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Why high-fidelity and why now Nuclear Science and Engineering Established Digital Twin

Advancement and ABWR demonstration of high-fidelity simulations based maintenance approaches and model based fault system detection techniques. Operating History Data Quantitative Data Assimilation Assessment for O&M

Address mechanical and BWRX300 thermal fatigue failure modes which drive O&M activities of BWRX-300, and are extendable to all advanced reactors (ARs) where a flowing fluid is present.

High Fidelity Virtual Quantitative Operation Model Data Assimilation Assessment for O&M High Fidelity Digital Twin 138 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE why now Nuclear Science and Engineering Computational Fluid Dynamics is an old tool, it has been leveraged by various industries for many years, and it has provided great support to design and safety of nuclear reactors as a complement to experimental and experience based approaches CFD has a much greater potential: to provide high-fidelity data to support efficient operation of NPPs bootstrapping the lack of operational data The incomplete maturity of the simulation methods (and teams) and the excessive computational costs has hindered this last major jump.

Leveraging the last 10 years of DOE and industry sponsored effort we are in position to demonstrate this jump 139 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE why now Nuclear Science and Engineering Computational Fluid Dynamics is an old tool, it has been leveraged by various industries for many year, and it has provided great support to design and safety of nuclear reactors as a complement to experimental and experience based approaches But CFD has a much greater potential: to provide high-fidelity data to support efficient operation of NPPs bootstrapping the lack of operational data The incomplete maturity of the simulation methods (and teams) and the excessive computational costs has hindered this last major jump.

T-junction blind international benchmark 2013 had good and bad news

LES works very well, but its too slow to drive Operational DTs

Many acceleration ideas proposed for external aero, some successful, but not for Nuclear Applications For Nuclear Applications we need robust reliability:

We are looking for local hybridization in presence of turbulent structures

Independence from grid resolution

Independence from time stepping and spatial interpolation methods Leveraging the last 10 years of DOE and industry sponsored effort we are in position to deliver this capability 140 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Thermal striping driven fatigue prediction Nuclear Science and Engineering Thermal striping fatigue has largest applicability to all AR concepts, beyond BWRX-300 - FIRST OBJECTIVE A few notorious examples 141 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE The challenge in a (hand drawn) nutshell Nuclear Science and Engineering

We can leverage the Civaux failure for discussion 180 mm 1Hz 1Hz

The turbulent structures generated at the 90°elbow interact at the T and lead to low frequency (1-3 Hz) large temperature oscillation that lead to accelerated fatigue failure

The same T connection without the elbow does not suffer of these oscillations (but small flow / geometry variations could lead to the opposite results)

Phenomenon is driven by formation and interaction of large turbulent structures and is strongly non-linear, not prone to lumping and generalization 142 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE The STRUCT idea Nuclear Science (Lenci and Baglietto, 2016) and Engineering In regions with separation, jets, Scale overlap swirls, and strong mixingflow deviates from equilibrium Velocity Time URANS is not applicable to scale overlap (U)RANS models are based Velocity on the assumption of an equilibrium spectrum High frequency part Low frequency part Time http://ffden-2.phys.uaf.edu/647fall2013_web.dir/j_stroh/tec.html 143 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE The STRUCT idea Nuclear Science (Lenci and Baglietto, 2016) and Engineering STRUCT has demonstrated consistent grid convergence and accelerations between 50-100x on a number of validations Thanks to DOE and Industrial sponsorship Supported by:

144 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Recap: coherent flow structures drive T variation Nuclear Science and Engineering Maturity of models and application experience, at reduced computational cost URANS (Cubic k-) STRUCT-URANS (Cubic k-) STRUCT-Mean Temperature Var Temperature URANS Cannot predict Velocity STRUCT predicts both and Temperature Fluctuations velocity and Temperature responsible for fatigue fluctuations with high accuracy Inst. Temperature Errors on Trms order of 1%

Resolved Coherent Temperature Structures 145 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Application on feedwater system recap Nuclear Science and Engineering Manchester LES STRUCT Velocity Upper part vortices Vector Lower part vortices A method for flow feature extraction POD Mode 1

- The STRUCT model accurately captures the swirl switching features 146 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Swirl-Switching example Nuclear Science and Engineering Fluid (~3.5 inch downstream)

100%

50%

147 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE High-fidelity simulation based DT generation Nuclear Science and Engineering The model outputs will look like this:

Labels Point 1s stress Point 2s stress Point ns stress (BC) time-series time-series time-series 1st operation conditions 1 1 1 1 1,1 1 1,1 1, 1 1, ( ) 1, 1 1, ( )

2nd operation conditions 2 2 2 2 2,1 1 2,1 2, 1 2, ( ) 2, 1 2, ( )

The high-fidelity data provide a uniquely rich database for DT generation Capability of including sensitivity to many parameter variations 148 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE High fidelity Data Collection Nuclear Science and Engineering Origin Probes:

= . ~

= ~

675 points at the inner wall ( = 7.625 )

675 points at the outer wall ( = 8 )

149 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

NSE Digital Twin Prototype Nuclear Science and Engineering Feedwater subsystem Analysis of number of cycles to damage initiation ( ) Uncertainty Quantification in Crack Initiation using simulated data at each power level:

Power Level

1. Selection of power level (user input) 20% 100%

2. Running Rainflow Counting on the available Cycles simulated stress histories

3. Calculating values 1st Cycle

4. Modeling across all angles and positions in the . . .

data for each cycle using Gaussian Process models Rainflow Counting Algorithm Crack initiation Predictions across all positions and angles .. ..

Uncertainty Quantification in crack initiation nth Cycle Identifying number of cycles in stress . . .

histories Smaller

~ x 150 MIT - HIGH FIDELITY DIGITAL TWINS FOR BWRX-300 CRITICAL SYSTEMS September 14-16, 2021

Molten salt-loop development acceleration with distributed single-crystal harsh-environment optical fiber-sensors Presented by: Michael Buric, Staff Scientist NETL 2019-2022 ARPA-e TINA, 2021 update

Crystal fiber distributed sensing Project Objectives

Introducing fully-distributed sensing to Molten-Salt Reactors

Growing new cladded single-crystal optical fibers for molten-salt environments

Gathering thousands of data-points to map reactor coolant-path temperatures or other parameters

Mapping in-core temperature distributions

Next-gen sensing replaces single-point sensors like thermocouples

Providing data to guide reactor design and improvement through thermal efficiency

Support LFMSR Licensing Basis 152

Crystal fiber distributed sensing Team National Energy Technology Lab (fiber growth, sensor design, interrogator design)

- Michael Buric (PI, fiber optics and systems)

- Guensik Lim (LHPG)

- Juddha Thapa (LHPG, materials)

- Jeff Wuenschell (DTS, testing)

Idaho National Lab (reactor expertise, system implementation and testing)

- Pattrick Calderoni (in-pile instrumentation director, co-PI)

- Joshua Daw (nuclear instrumentation)

- Ruchi Gakhar (nuclear materials)

MIT (material compatibility, efficacy simulations)

- David Carpenter (Irradiation Engineering Director)

- Koroush Shirvan (reactor design and simulation, co-PI)

- Tony Zheng, Yeongshin Jeong 153

GOAL: MSR Thermal Response Analysis for sensor specification and placement

Providing information on the transient responses under possible transients of molten salt reactor for monitoring system with fiber optic sensor Neutronics Generation of neutronics parameters (SERPENT) 0-D/3-D, Steady state Neutronics: static fuel at constant temperature Coupled Transient response of state variables (power, salt and neutronics/T-H structure temperature, reactivity, etc.)

(Simplified MSR Kinetics Model)

Lumped model, transient P Neutronics: effects of precursor transport due to flow T/H: energy balance only, constant flow rate t

Thermal-Hydraulics Thermal response of MSR in time and space (Star-CCM+) 2-D (or 3-D), transient T1 Neutronics: power changes imported from simulator T2 t T/H: momentum/energy balance incl. convective and t

TN radiative heat transfer of salt t

Simple System Transient Model Completed MCFR Application (Based on Available Literature Validation on TerraPower MCFR Design)

Step reactivity insertion (+2)

ORNL MSRE Data Verification European MSR Simulator C. TRIPODO et al.,

EPJ Nuclear. Sci.

Technol., 5, 13 (2019). =+2 Transient response of power, reactivity, and fuel salt temperature changes (1)

3D Accident Analysis Completed

Demonstrates the Techno-Economic Value of Continuous Temperature Monitoring Temperature distribution in the salt, reflector and vessel walls experience meaningful gradients and local peaks Unprotected LOF (Decrease of fuel salt flow rate to 80 %

exponentially with time constant of 5 sec)

3D Accident Analysis Unprotected LOF 7

6 5

4 T1 2 3 R7 R6 R5 C1 R4 R3 R2 B1 2 3 R1 4

5 6

7

Crystal fiber distributed sensing - LHPG How LHPG works:

CO2 laser melts oxide feedstock

Seed crystal lowered into melt

Controlled motion of seed and feedstock upward

Fiber is grown from the melt 125 µm Fiber Melting pool Pedestal 158

Crystal fiber distributed sensing Grow cladded fibers with 2-stage LHPG

- Sapphire or YAG

- Sol-gel (or other) dopant additions Evaluate materials compatibility in fluoride and chloride salts (bench tests)

Evaluate radiation durability (gamma source, research reactor)

Grow pure fibers Prepare sol- Coat pure in YAG, Sapphire, gel-based fibers via dip-etc. (LHPG #1) dopants coating TEST clad fibers Re-grow coated in application fibers (LHPG environment #2) 159

Crystal fiber distributed sensing - Claddings 1.95 1.95 Holmium Holmium 1.75 1.75 Neodymium Neodymium 1.55 1.55 1.35 1.35 1.15 1.15 0.95 0.95 0.75 0.75 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Automatic Dopant Segregation through LHPG: Top left: Visible light guiding in GRIN YAG fiber, Top right: EMPA map of Nd concentration in a GRIN YAG fiber, Bottom plots: Co-doped Nd and Ho: YAG fiber dopant concentrations in X (left) and Y (right) 160

Regrowth LHPG System 161

Distributed Sensing - Raman OTDR OTDR with Single Crystal fiber Useful for high-rad/ high-T

~5cm, ~1C resolution Sapphire fiber attenuation at Raman OTDR, Liu et al Opt. Lett., 2016 532nm, measured by a Raman Attenuator Lens 1 OTDR system 0

PD 1 Laser clean-up filter Beam 50:50 Beam Lens 2 damper splitter -5 Sapphire fiber Pulsed laser Normalized Log scale intensity (dB)

Syn.

-10 Stop line filter Stokes filter Lens 3 -15 Dichroic beam splitter APD 1

-20 OSC PC Anti-Stokes filter Lens 4

-25 Mirror APD 2 0.5 1 1.5 2 2.5 Distance (m) 162

Raman DTS design and operations Current view of Raman DTS in the lab with external test systems

DTS design for portability / finished product prototype

Flight case design

Laser safety - electrical interlocks

Software for lead-in fiber

Field tests in July at INL

(even more important) field test in October at MITRR 164

Furnace calibration with 50 standoff New data processing for long-standoff

With 50-ft. lead fiber

Apply small delay in Stokes vs. Anti-Stokes based on reflection peak delay (one light-ns is 300mm)

Calibrating only on range where temperature-based losses are minimal (300-700 C).

Lower loss fiber leads to higher Temp capability

Smaller fiber leads to higher temp capability

INL Molten salt field tests Testing up to ~750C Molten Chlorides Assortment of crystal fiber materials and protection layers 167 167

MITR test planning Encapsulation tubing (pressure boundary) for fiber dummy fuel element insertion DTS standoff at ~60 feet October 11th installation Dry irradiation vehicle Dummy fuel element Gas/instrumentation tubes 168

Conclusions Distributed sensing is coming to numerous industries Single-crystal Optical fiber technology can extend into nuclear harsh-environments Raman DTS is a good distributed platform for SC-fiber Temperature mapping needed for LFMSR transient response Amazing new levels of visibility and automation are here!

169

Digital Twins to Production Reactors The Simulation Continuum Bob Urberger, Roger Chin 09/15/2021

Who Are We? l Radiant

Radiant was founded by former SpaceX engineers in 2019 to make nuclear power portable.

Currently designing and building Kaleidos, a 1 MW electrical reactor that fits in an ISO shipping container.

We believe our experience from the aerospace and software industries will allow us to bring a safe, reliable product to market quickly.

171

Digital Twins as a Tool Digital Twin Regulator Developer Parameters and Control Systems/Realtime Documentation Source Modeling & Simulation Verification and Validation Anomaly Response Modeling High Fidelity Modeling & Simulation Performance Assessment Real-Time Modeling & Simulation Machine Learning Applications Digital Twins are a common tool between regulators and developers to ensure common sources of information for aspects relevant to them. 172

Why Do We Simulate?

Inform Design Aspects

Test Operational Procedures

Iterate Quickly and Inexpensively

Real systems are always correct and used for validation

Simulation purpose must be explicitly defined and designed for Simulations accelerate design and build confidence in a real system by informing decisions. 173

The Radiant Simulation Continuum

Our digital twin is low fidelity. How do we bridge the gaps?

Exploit the strengths of other simulation technologies.

The scope of each simulation must be carefully defined.

What does this sim model?

How accurate is it?

What can it NOT model?

Digital NEAMS Hardware-In- Prototype Digital Twin Hybrid Sim Tools The-Loop Sim Systems Physical Digital Twins fill a missing role in between high fidelity simulations and physical hardware 174

Digital Twins Strengths Digital Twins Digital Twins in Context Control Systems Digital Twins NEAMS Hybrid HIL Prototype 4

Control Systems 3 5 Digital PRA 2 Hardware Design 4 3

1 Digital PRA Hardware Design 2

0 1 0

Costs FMEA and Risk Costs FMEA and Risk Manufacturability Manufacturability Digital NEAMS Hardware-In- Prototype Digital Twin Hybrid Sim Tools The-Loop Sim Systems Physical Digital Twins are complementary to existing technologies for a development cycle.

The Radiant Simulation Continuum

SimEngine is Radiants in-house simulation tool.

It can run in real time.

It can run with any mix of real and simulated hardware

Each model can be anchored to either a higher fidelity digital model or a higher fidelity physical model

Real hardware is used as the source of truth when possible.

NEAMS tool solutions used as source of truth elsewhere.

Digital SimEngine NEAMS Hardware-In- Prototype Digital Twin Hybrid Sim Tools The-Loop Sim Systems Physical SimEngine is Radiants In-House, Real-time Digital Twin 176

Summary

Reality is the ultimate truth. Digital twins have limitations.

Simulations must have a defined scope.

What can/cant they predict? How accurate are they?

Simulations must be anchored.

How do we know the simulation is correct? We need to know the source of truth and the simulations relationship to it.

A single high-fidelity digital sim isnt practical for Radiants use. Instead, we cover a larger simulation space with multiple narrowly focused simulations.

177

NSE Nuclear Science and Engineering Thank you for your attention 178

Steps Toward Regulatory Realization of Digital Twins Closing Panel Session - September 16, 2021 Moderator: Eric Benner, NRC Jeremy Bowen, NRC Tom Braudt & Steve Vaughn, X-energy Paul Keutelian, Radiant James Slider, NEI Anthonie Cilliers, Kairos Power Brian Golchert, Westinghouse

Digital Twins -

Regulatory Viability Jeremy Bowen The 2021 Workshop on Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 14 - 16, 2021

Digital Twins Project Plan PREPAREDNESS TECHNICAL REGULATORY PREPAREDNESS READINESS FFR BL ASSESSMENT OF COMMUNICATION PROJECT FUNDED STANDARDS & KNOWLEDGE MANAGEMENT

Takeaways Thus Far The State of Technology of Application of Digital Public Workshop #1 Twins Assessment of the state of DT Widespread interest with technology in nuclear reactor >400 participants across applications the globe Technique is increasing Proven benefits and developing rapidly Development of a common Need for community of understanding practice to collaborate Access Link: ML21160A074 Access Link: ML21083A132

Active & Future Tasks Technical Challenges and Gaps Technical Preparedness -

for Digital Twins in Using Data Safeguards and Security in Analytics, ML/AI and Digital Twins, Multi-Physics Models Regulatory Readiness Levels and Online Monitoring for Gaps Pertaining to Digital Twin Enhanced Diagnostics Technologies for Nuclear and Prognostics Reactor Applications

Steps Toward Regulatory Realization of Digital Twins Xe-100 Licensing Perspectives Tom Braudt and Steve Vaughn, X-energy Licensing September 16, 2021

Xe-100 Digital Twin Tools 3D Models with AR / VR Operator Training Simulator Plant Historian AI / ML Models

Hi-Fidelity Simulator and 3D Models Simulator Certification Simulator HFE Validation

Regulatory Intersections with the Xe-100 Digital Twin Digital Twin Primary Purpose is Plant Economics - Not Regulated Some Intersectionality with:

Xe-100: extensive use of digital control and monitoring systems

Maintenance Rule - predictive maintenance

Simulator fidelity and configuration management

Human Factors Engineering - predictive trending and intelligent alarms reduce operator workload

NLO/Maintenance/I&C/RP Training - multi-role approach

ALARA dose reduction

Security analytics and what-if scenarios

Fire Detection - intelligent dispatch and response

Support load-follow, industrial heat, hydrogen production Strong NRC Staff Interest in Understanding Xe-100 Digital Twin Capabilities and Usage

Control Room - staffing methodology

Operator Training - methodology and workload

Digital Twin Regulator Pathway - Future Considerations Use of AI / ML for Control Functions

Ensuring regulatory framework positioned to support

Leverage lessons-learned from safety-related digital retrofits (cyber)

NUREG/CR-7273 starting point Dynamic PRA Usage - initiating event prediction via AI Pre-emptive utilization of Remaining Useful Life (RUL) for equipment Sensor Drift Detection Cyber Security Requirements for Control Functions - DoD Lessons Learned Encourage Research at NR/ INL/ORNL to Support Control Functions Making digital twins to monitor and control tomorrows reactor designs - INL

A Personal Perspective on Digital Twin Capabilities: no 2-seat F-35s

Using Digital Twins to Support Regulations Paul Keutelian September 16, 2021

Vision l Radiant Make Nuclear Portable Maximize speed of iteration to get to a buildable product Provide streamlined regulatory analysis 13

Novel Challenges Regulatory Intents Protect Personnel Protect Environment Digital Twins ?

Protect Hardware Prove the Above Digital Twins are a powerful potential to answer these core questions, the challenge comes from understanding where is it appropriate to use this tool, and how do we work with regulators to build confidence? 14

Digital Twins as a Tool Digital Twin Regulator Developer Parameters and Control Systems/Realtime Documentation Source Modeling & Simulation Verification and Validation Anomaly Response Modeling High Fidelity Modeling & Simulation Performance Assessment Real-Time Modeling & Simulation Machine Learning Applications Digital Twins are a common tool between regulators and developers to ensure common sources of information for aspects relevant to them. 15

Contributions

Internal analysis documentation standardization

Visualizations for eased introduction to analysis

Automation of explicit analysis flows

Common interface with Regulators 16

JAMES SLIDER TECHNICAL ADVISOR

ANTHONIE CILLIERS SENIOR MANAGER INSTRUMENTATION, CONTROLS AND ELECTRICAL

BRIAN GOLCHERT PRINCIPAL ENGINEER

Enabling Technologies for Digital Twin Applications for Advanced Reactors and Plant Modernization September 14-16, 2021 Thank you for your participation in this workshop. The proceedings for the workshop will be publicly available in the next few months. Please provide any comments or feedback to one of these workshop sponsors.

Jesse Carlson, NRC - jesse.carlson@nrc.gov Vaibhav Yadav, INL - vaibhav.yadav@inl.gov Jenifer Shafer, ARPA-E - jenifer.shafer@hq.doe.gov Hasan Charkas, EPRI - hcharkas@epri.com Prashant Jain, ORNL - jainpk@ornl.gov

ML21348A020

Contents

Text

SUMMARY

SUMMARY

2.1 Session

2.1 Session

Contact:

Contact:

Contact:

Contact:

Contact:

SUMMARY

Background:

References:

Background:

Title:

Navigation menu