Why Most Plants Are More Automated Than Autonomous

KEYWORDS: Autonomous Operations, Industrial Automation, Decision Support, Process Industries, Industrial AI, Operational Technology

Overview

The first article in this series argued that autonomous operations is not primarily about removing people from industrial operations. It is about applying human expertise where it creates the greatest value. This second article builds on that idea by examining a related issue: many plants described as autonomous remain primarily automated. They use advanced control systems, historians, dashboards, alarms, predictive models, and digital workflows, yet critical decisions still depend on people to interpret information, select responses, and coordinate action.

This distinction matters because the road to autonomous operations is not only a workforce journey; it is also an authority journey. Automation executes predefined actions. Decision support helps people make better decisions while decision rights remain with them. Autonomy requires systems to receive delegated authority to select and execute actions within approved operating boundaries, escalate exceptions, and adapt as conditions change. Automation executes. Autonomy authorizes. The transition requires changes in operating models, decision rights, workforce roles, data architecture, cybersecurity, and the operational signals needed to support the authorization of actions.

Plants can be highly automated and still far from autonomous if humans must continuously interpret, approve, coordinate, and execute nonroutine responses.

For process industry operators, the practical question is not whether a facility has significant automation. Most do. The more important question is whether automation has changed who or what makes decisions, how those decisions are executed, how exceptions are handled, and how accountability is maintained.

This Insight examines why most plants are more automated than autonomous, what has changed to make higher levels of autonomy more achievable, and what owner-operators should do to move from automated execution and decision support toward autonomous operations with clearly delegated authority. It also sets up the next step in the series: the hidden role of alarm management in providing the operational context needed to authorize actions.

Most Process Plants Are Already Highly Automated

Process industries have invested in automation for decades. DCSs, PLCs, SISs, SCADA, MES, APC, alarm management, historians, and asset monitoring tools are common across refineries, chemical plants, LNG facilities, pipelines, terminals, power generation assets, mining operations, and other complex industrial environments.

These systems deliver enormous value. They improve consistency, maintain operations within defined limits, reduce manual manipulation, support safer operation, and give operators visibility into plant conditions. They also form the foundation for higher levels of autonomy.

However, automation does not equal autonomy. Many plants still operate with legacy equipment, standalone applications, point-to-point integrations, manual procedures, and human-led coordination. The technology inventory may look advanced, while the operating model remains fragmented.

The result is a facility that appears advanced but remains dependent on human interpretation and execution. Control systems maintain setpoints, historians capture data, alarms notify operators, analytics identify anomalies, and maintenance systems generate work orders. Yet the chain from detection to decision to action often remains human-centered. This is why “we have automation” should not be confused with “we have autonomous operations.” Automation can reduce manual tasks, improve visibility, and generate recommendations without reallocating decision rights or granting systems the authority to execute actions.

For technical managers, this distinction determines where investments should go next. A plant with mature automation may still need better context, orchestration, procedural automation, cybersecurity, workforce acceptance, and operating model design before it can safely delegate decision and execution authority. The quality of operational signals also matters because autonomy cannot authorize appropriate actions when those signals are dominated by noise.
 

ARC Advisory Group clients can view the complete report at the ARC Client Portal.

Contact Us if you would like to speak with the author.

Obtain more ARC In-depth Research Market Analysis.