MTA-EN-015

Return to MTA Table

Administrative Items
Date	2/1/2025
Functional Area Where Benefits Will Be Realized	Engineering Performance Improvement Operations Maintenance
Reference Implementation Guidance	The technology is relatively new and vendor‑driven, therefore no general industry guidance is available to date. A technology assessment report will be published soon by the IAEA. See documentation linked below on an open‑source, free Modelica modeling library for the development of one Digital Twin solution. Example Modeling Library
Industry SME	EPRI PRR Contact: NuclearPlantMod@epri.com
Previous Implementation	This technology has been implemented at several nuclear and gas plants. Please contact the EPRI SME for additional information.
Implementation Enablers	N/A. The technology is mature and immediately available. It requires adequate sensor coverage for the selected plant scope, which is practically the case for most power conversion cycles.
Applicability	All reactor types All geographic regions
Keywords	Monitoring; Diagnostics; Automated Fault Detection; Digital Twin; Performance; Efficiency; Maintenance; M&D.
Business Case Analysis Cross-Reference	EPRI is considering a new BCAM for use of Digital Twin to improve plant performance monitoring. There are several existing BCAMs that discuss this topic using alternate technologies: Plant Modernization Business Case‑Improved Thermal Performance Through Cycle Isolation Monitoring: Cost‑Benefit Analysis of Cycle Isolation Monitoring for Addressing Valve Repairs That Lead to Lost MWe (EPRI 3002019844) Plant Modernization Business Case: Improved Thermal Performance Through Data Validation and Reconciliation: Cost‑Benefit Analysis of DVR Analysis for Power Recovery or Measurement Uncertainty Recapture Uprates (EPRI 3002019845) Plant Modernization Business Case: Monitoring and Diagnostic (M&D) Program Development: Cost‑Benefit Analysis of Implementing Online Monitoring (OLM) and an M&D Program to Reduce Operating Costs and Avoid Outage and Derate Costs (EPRI 3002028178)

Description

Power conversion cycles in aging light‑water reactors typically run with 3‑5 simultaneously active faults, leading to significant power losses. Those faults may include heat exchanger fouling or tube ruptures, leaking steam isolation valves, turbine/pump underperformance, and sensor bias issues. Diagnosing those faults promptly can be challenging, particularly when their impact on sensor readings is minimal, distributed across numerous sensors, or includes overlapping contributions with varying positive or negative effects from the active faults. Technologies increasing the accuracy and speed of fault diagnostics can provide substantial economic benefits to the plant, including rapid recovery of lost MWe and avoidance of the cost of some consequential faults through early detection.

Key benefits are obtained through Digital Twin fault inference technology. The technology supports monitoring and diagnostics (M&D) plant staff and is best described in two steps: quantification of 1) symptoms, and then 2) faults. Technologies that address symptoms without faults are also available.

1. Symptoms, defined as deviations of actual versus healthy expected sensor readings, are typically quantified by comparing measurements from the real plant with virtual measurements performed in a Digital Twin. The Digital Twin is a physical model fed with real‑time boundary conditions. Validity of this step depends on how accurate the Digital Twin is at replicating a realistic, healthy performance state of the plant as well as how optimally sensors are placed in the system to capture the needed physical quantities with adequate accuracy. For example, calibrated models tend to describe real plant conditions more closely than reference (nameplate) models.
2. Faults, hereby defined as issues with an impact on pressure, temperature, and/or flow measurements that affect the performance and/or reliability of the system. Faults are identified and quantified by feeding the symptoms from step 1 above to an inference algorithm that is also provided with a library describing the signature of individual postulated faults on symptoms. The ability to detect simultaneous faults depends on the accuracy and completeness of the fault library informing the inference algorithm and on the performance of the algorithm at successfully identifying and quantifying complex combinations of simultaneous faults. Methods are available to build the library and include literature documents (e.g., the EPRI Thermal Performance Engineering Handbook, Volume 3 – (EPRI 3002005346) troubleshooting guide), fault tree analyses, or fault signatures that are explicitly modeled using the Digital Twin.

Benefits

Benefits Estimate

Level 1 – Savings are less than $1 M per year.

More specifically, savings for a typical ~1 GWe unit’s steam cycle are estimated at about $600 K per unit‑year. About two thirds of those savings come from increased energy production derived from prompt recovery of MWe losses. The remaining third comes from benefits of early problem detection (asset life extension, repair expense prevention, forced outage prevention) and increased workflow efficiency (inspection, M&D, training, communication).

Benefits Description

• Increased energy production from rapid recovery of lost MW
• Increased energy production from increased reliability
• Improved maintenance planning
• Extended asset life from early detection (example: heat exchanger tube integrity)
• Reduced repair costs from early detection
• Improved cycle isolation leak management
• Consistent knowledge of plant state, easily communicated
• Improved personnel training and plant history reporting
• Shared fault knowledge at plant and fleet level

Costs and Schedule

Cost

Level 3 – Implementation cost is less than $1 million.

The cost of Digital Twin fault inference technology depends on the vendor and solution adopted and is typically broken down into two key parts:

• A one‑time modeling cost to build the Digital‑Twin‑based implementation at a specific plant unit. This cost decreases for added units of same or similar design.
• Software licensing cost to run the real‑time fault inference algorithm; this can be offered in Software‑as‑a‑Service (SaaS) mode including platform and model maintenance lumped into a fixed fee.

As the Digital Twin fault inference technology is not compromised by the emergence of new faults, being able to quantify them, model maintenance efforts are not frequently needed. They would be needed when the system or its setpoints are modified.

The technology runs on existing plant sensors, so it does not require hardware costs.

Schedule

Six months to one year. The detailed schedule depends on the vendor and solution. Usually, building a Digital‑Twin‑based implementation may take about six months, which includes carrying out preparatory work to establish reliable live data workflow.

Scope Context

A single system or a group of systems within a unit. Depending on the vendor and solution, the technology can be expanded to monitor groups of units or fleets. The system scope is typically for steam cycles but extension to other systems is possible.

Risks

The following risks may materialize depending on the technology and deployment specifics:

• False‑positive and false‑negative fault detections are possible.
• Lack of diagnostics accuracy may lead to potential failure in obtaining a return on investment.
• Typical implementation delays from IT, legal, and export control approvals for live data flow may occur.
• Change management difficulties may occur in adoption and use in detection and remediation workflows.

SWEEP Score

Category	Level	Description
Cost	3	Implementation cost is less than $1 million
Savings	1	Savings are less than $1 million per year
Payback	3	Payback period is less than one year
Licensing Readiness	3	No changes are required for implementation
Technology Readiness	3	The technology is ready for wide operational deployment
Implementation Proficiency	3	The technology can be implemented by all sites, regardless of digital experience