--- 
authors:

- "Alberti, Anthony L."
  
  - "Agarwal, Vivek"
  
  - "Gutowska, Izabela"
  
  - "Palmer, Camille J."
  
  - "de Oliveira, Cassiano R. E."
  
citekey: "albertiAutomationLevelsNuclear2023" 
alias: "albertiAutomationLevelsNuclear2023" 
publish_date:
2023-03-01 journal: "Progress in Nuclear Energy" 
volume: 157
pages: 104559 
last_import: 2025-07-30 
---

# Automation levels for nuclear reactor operations: A revised perspective

## Indexing Information Published: 2023-03

**DOI**
[10.1016/j.pnucene.2022.104559](https://doi.org/10.1016/j.pnucene.2022.104559)
#Small-modular-reactor, #Microreactor, #Advanced-sensors,
#Artificial-intelligence, #Automation-levels, #Digital-twin,
#Fission-battery, #Reduced-order-model


#InFirstPaks



>[!Abstract] This work serves to propose updated levels of
>automation for nuclear reactor operations, as a result of
>considering long-term economic and commercial ambitions of
>the advanced reactor developer community. As in other
>fields such as road-going vehicles and aviation, reactor
>technologies can benefit from modern automation through the
>resulting reduction in operations and maintenance costs,
>while still maintaining the current industry standards
>regarding safety, resilience, reliability, overall
>performance, and the capacity for root-cause analysis. The
>current guidelines on automation levels, as published by
>the U.S. Nuclear Regulatory Commission in Section 9 of
>NUREG-0700, reflect outdated design principles that
>implicitly limit the potential of automation innovation for
>reactor operations, particularly in regard to advanced
>reactors intended to operate in remote locations or be used
>for off-grid applications. Motivated by the operational
>paradigms anticipated for future reactor designs, we
>propose a six-level approach that aligns with contemporary
>automation concepts as well as automation level definitions
>from other non-nuclear safety–critical industries. These
>levels build upon the current guidelines in order to enable
>next-generation nuclear reactor technologies to become
>increasingly economically competitive and commercially
>viable relative to competing power generation sources.
>Using a hypothetical heat removal reactor transient, we
>provide examples of how the human–machine interactions
>change at each level of automation, ranging from
>traditional operator control (Level 0) to operator-free
>unattended operations (Level 5)—the latter being one of the
>key attributes proposed at the Fission Battery Initiative
>led by Idaho National Laboratory. Finally, we critically
>examine the identified challenges, knowledge gaps, and
>enabling technologies to achieve advanced levels of
>automation.>[!seealso] Related Papers
>

## Annotations ### Notes ![[Zettelkasten/Literature
Notes/Notes on Papers/Automation levels for nuclear reactor
operations- A revised perspective-notes.md]]

### Highlights From Zotero
>[!highlight] Highlight we propose a six-level approach that
>aligns with contemporary automation concepts as well as
>automation level definitions from other nonnuclear
>safety-critical industries. 2025-07-24 9:42 am
>

>[!done] Important For typical reactors in the current
>fleet, 70% of their non-fuel O&M costs relate to labor
>[12]. 2025-07-24 9:44 am
>

>[!tip] Brilliant The various levels of automation, as per
>Section 9 of NUREG-0700, ultimately rely on a
>human-in-the-loop to monitor plant performance and
>intervene when necessary. Under these guidelines,
>automation technology aims to provide operator support,
>rather than replace operator duties in regard to sustained
>day-to-day operational and tactical control. 2025-07-24
>9:51 am
>

>[!tip] Brilliant Recognizing that operations plans
>contradicting these guidelines can receive NRC approval
>through sufficient reasoning, we posit that this
>requirement implicitly limits the ambitions and desired
>operational paradigms of the advanced reactor developer
>community, and creates a potentially detrimental disconnect
>with the regulator. This is particularly applicable in the
>case of fully autonomous human-out-of-the-loop-based
>operations (a key facet of the “unattended” attribute
>adopted in the FB Initiative [21]). 2025-07-24 9:51 am
>

>[!done] Important In terms of classifying tasks for
>automation in nuclear power plant operations, O’Hara and
>Higgins break down the tasks for which reactor operators
>are responsible into two general types: primary and
>secondary [23]. Primary tasks, which are the main focus of
>this work, directly impact the overall functionality and
>safety of the plant. Secondary tasks encompass all other
>intermediate tasks, such as navigating, arranging, and
>interrogating information at workstations and control
>panels [24, 23]. 2025-07-24 11:22 am
>

>[!tip] Brilliant Monitoring and detection covers activities
>related to obtaining information about a system or
>subsystem. This can simply mean tracking parameters on a
>control panel or sending personnel to visually verify that
>a component is functioning properly. Situation assessment
>refers to evaluating obtained data and determining the
>state of a system or subsystem. This typically involves
>understanding whether the given plant or system is
>operating properly. If any anomaly is found, the underlying
>causes are investigated. Response planning refers to
>determining proper courses of action, based on the
>situation assessment. Response implementation encompasses
>performing the actions identified in the response plan.
>2025-07-24 11:29 am
>
> *This is all really important, and outlines the chain of
> actions that a nuclear plant operator would take.*



>[!highlight] Highlight Strategic operations consider the
>questions of whether, when, and where. 2025-07-28 11:38 am
>

>[!highlight] Highlight Operational and tactical operations
>pertain to the context of vehicle motion. These impact
>longitudinal (acceleration/deceleration) and lateral
>(steering) actions, object detection, maneuver planning,
>and response execution. The number/combination of automated
>driving tasks—as well as which operational design domain
>they fall underlargely determine the automation level of a
>particular driving task. 2025-07-28 11:38 am
>

>[!highlight] Highlight The minimal risk condition is a
>predetermined state in which either a user or automated
>system recognizes a potentially hazardous situation and
>subsequently circumnavigates it to minimize risk. Fallback
>is the response to encountering a potentially hazardous
>situation 2025-07-28 11:43 am
>

>[!tip] Brilliant Level 3: The system recognizes a crash
>scene and requests that the driver resume control and
>provide fallback (e.g., engage hazard lights and enter the
>unobstructed shoulder lane).  Level 4/5: Even if the driver
>is unresponsive to the fallback request, the automated
>system is able to achieve the minimal risk condition or
>circumnavigate the hazard. 2025-07-28 11:44 am
>

>[!done] Important The automation tipping point occurs at
>Level 4, with the automated flight system becoming able to
>sufficiently control operational and tactical tasks, such
>that a human pilot is no longer required. Rather, the ECA
>suggests the implementation of a mission commander, who is
>in command but not in control. 2025-07-28 11:48 am
>
> *Higher levels of optimization actually reduce the
> training demand of operators. We see this all over the
> place with things like ChatGPT. They remove the grunt work
> so the human can think **strategically**.*


>[!tip] Brilliant At Level 5, the commander simply provides
>strategic commands to the system—without controlling the
>aircraft at any point. 2025-07-28 11:48 am
>

>[!done] Important Additionally, we assume strategic
>operations to not be automated at any level. These are
>reserved for human command, either onsite via an operator
>or reactor supervisor, or offsite via a reactor supervisor.
>Only operational and tactical tasks are considered for
>automation. 2025-07-28 11:53 am
>

>[!highlight] Highlight Following the definition presented
>by the SAE, automatic active safety systems (e.g., the ADS
>in the ESBWR) are not classified as automated. Rather,
>these systems provide safety measures to ensure the
>ultimate safety of the plant, and thus should be included
>at every automation level. 2025-07-28 12:59 pm
>



### Follow-Ups

>[!example] In the present paradigm, much of this automation
>comes in the form of operator support. Examples include
>computerized operator support systems that assess various
>alarms and provide fault diagnoses to operators [26, 27,
>28, 29], computer-based procedure systems that provide
>necessary data and procedures to assist operators by
>identifying tasks in real-time to foster safety goal
>achievement [30, 31, 32], and the automatic activation of
>primary safety systems (e.g., emergency core cooling
>systems) during severe accidents.
>- [ ]  #Follow-Up
  
>[!example] [23] J.M. OHara and J.C. Higgins. Human-system
>Interfaces to Automatic Systems: Review Guidance and
>Technical Basis. Technical Report BNL–91017-2010, 1013461,
>2010. URL https://doi.org/10.2172/1013461. [24] John M.
>O’Hara, William S. Brown, Paul M. Lewis, and J.J.
>Persensky. The Effects of Interface Management Tasks on
>Crew Performance and Safety in Complex, Computer-Based
>Systems: Overview and Main Findings (NUREG/CR-6690), 2002.
>URL https://www.nrc.gov/
>reading-rm/doc-collections/nuregs/contract/cr6690/vol1/index.html.
>- [ ]  #Follow-Up