Reliability and maintenance leaders have spent years chasing uptime with preventive schedules and manual investigation. Industrial artificial intelligence is now shifting that work toward prediction, faster action, and reliability programs that scale across sites.

Why It Matters

Unplanned downtime still drains production and inflates maintenance budgets across process manufacturing. In a recent episode of The Main Column Podcast, Nithiya Parameswaran, Vice President of Product Management at Emerson’s Aspen Technology business, joined Lee Nichols, Editor-in-Chief of Hydrocarbon Processing, to explain how Asset Performance Management (APM) is moving beyond monitoring toward predictive and prescriptive maintenance.

Data plus artificial intelligence (AI) can turn maintenance from a cost center into a source of better business outcomes. That change depends as much on discipline and change management as it does on technology.

Key Takeaways

• APM applies data to move maintenance from reactive and preventive toward predictive, prescriptive, and ultimately optimized plans.
• Combining instrumented asset data with failure history lets AI create failure signatures and detect anomalies before problems escalate.
• Agentic AI can now explain why an alert fired, cutting the manual investigation that reliability engineers used to shoulder.
• Successful programs start with available data and the right critical assets, not the most expensive ones.
• Templates and reusable libraries are what let a proven pilot scale to more asset classes and more sites.
• The next frontier includes corrosion, flaring, column flooding, and smarter capital and operating expense decisions.

How APM Evolved From Scheduled Maintenance to Predictive Insight

APM grew out of enterprise asset management (EAM) and computerized maintenance management systems (CMMS), which were built to schedule work, staff it, and stock the right parts. Those systems handled routine preventive maintenance well, but they treated all assets alike. Nithiya described how APM introduced criticality into the picture, matching maintenance intensity to an asset’s importance to production. From there, the discipline moved through condition monitoring and, once assets were instrumented, into data-driven predictive work.

Knowledge that once lived only in experienced people is now captured in the software as asset templates, failure mode and effects analysis (FMEA) libraries, and root cause analysis libraries, which puts hard-won expertise in front of reliability engineers when they need it.

Where Artificial Intelligence Changes the Work

AI earns its place in APM by pairing instrumented data with past failure history. Nithiya explained that this combination lets teams build failure signatures from prior events, then issue an advance warning when a familiar pattern repeats. That lead time is what gives maintenance crews room to act before a failure lands. AI also learns what normal operation looks like, so it can flag deviations as anomalies even when no known failure pattern exists.

The newest shift is interaction. Where predictions and anomaly alerts once left engineers to investigate on their own, agentic AI can now answer the question behind the alert. Nithiya offered a clear example: when pressure and temperature hold steady but flow rate changes, the system can point to what is actually wrong and explain its reasoning, doing much of the investigation for the engineer.

Breaking Out of Pilot Purgatory

Many APM pilots stall, and Nithiya was direct about why. Pilots fail when assets aren’t correctly identified or when teams haven’t confirmed their data is sufficient for the problem they are trying to solve. His guidance is grounded and repeatable: start at a single site, identify the critical assets that genuinely affect operations, confirm the data supports those use cases, and prove value there.

The most expensive asset is often not the culprit, because it is usually already well maintained. A $10,000 or $20,000 pump can be the one disrupting production. Once a program shows results, software is what makes scale possible. Templates and reusable libraries carry the expertise built for one site to other sites and asset classes, so teams expand on a foundation instead of starting over.

The Road Ahead: New Use Cases and Change Management

APM is reaching beyond failure prediction and anomaly detection. With process data alongside asset data, Nithiya sees room to manage corrosion, a persistent concern in oil and gas and around the crude distillation unit (CDU), by identifying causes and mitigation steps rather than only monitoring it. Fast-occurring events such as column flooding and flaring are also in view, and flaring carries both safety and feedstock-waste consequences.

APM can even inform investment planning, helping teams decide which assets to replace or extend. Nithiya closed with a candid point about discipline. Industry still runs preventive and predictive maintenance in parallel. He compared it to changing car oil every 3,000 miles when the vehicle already signals the right time. Capturing the full savings means revisiting preventive practices, and that requires deliberate change management, not technology alone.

Listen and Learn More

Hear the full conversation on The Main Column Podcast: How Industrial AI Is Transforming Asset Performance Management. To explore use cases, asset templates, and how Industrial AI is applied to reliability, visit the Asset Performance Management section on AspenTech.com.