How to Build a Self-Healing Plant

A future where plants will monitor, diagnose potential issues, and even schedule a fix is not far off – if we develop it correctly

Industrial manufacturers are under pressure. In oil and gas, for instance, margins are shrinking due to market instability, and ongoing skills shortages are threatening profitability in the medium to long term. Firms increasingly need to find ways to boost operational efficiency and source new revenue-driving opportunities just to remain competitive.

Developing an effective solution to this challenge is no easy task, but should be a priority – and addressing the issue of downtime needs to be a key strategic objective. The ongoing digitisation of industry is central to achieving this goal, primarily via the intelligent deployment of technologies like analytics, machine learning, automation, mobility, robotics, and AI. Integrated correctly, futurists envision a day when ‘self-healing’ plants will proactively monitor systems, diagnose potential issues, and even schedule a fix before they can lead to expensive unplanned downtime – all without human intervention. 

This isn’t just speculation – businesses and governments are already working together on ambitious projects to realise the self-healing plant. One such collaboration has seen the EU invest heavily in a consortium of 15 partners across six countries to improve health monitoring and life-long capability management for self-sustaining manufacturing systems.   

Starting with self-diagnosis

While it’s easy to speculate on how self-healing technologies will impact operational efficiency, the reality is they won’t come to fruition without the continued development of basic foundational elements. The most-important of these is arguably the ability to design and implement a self-diagnosing asset management system.

Representing a fundamental first step toward fully-automated approaches to plant maintenance, APM provides the data-driven insights required to detect problems before they escalate

Consider this example: a company determines that it requires 60 of its 100 wells to be running at any given time in order to meet output demands. If an equipment issue causes one well to fail, the system automatically knows to turn on another to ensure production requirements will not be interrupted and month-end targets can still be hit. Meanwhile, the system automatically detects the precise asset that failed during the operation, checks inventory, determines there are no spare parts, and then places a replacement order. Finally, it alerts engineers on site of the issue and the need to have the part replaced.

This is a simplified example of the role that asset performance management (APM) technologies will play in the facilitation of the self-healing plant. Representing a fundamental first step toward fully-automated approaches to plant maintenance, APM provides the data-driven insights required to detect problems before they escalate. However, traditional APM requires some significant tweaking to achieve the self-diagnostic capabilities required for a true self-healing plant.

Self-healing depends on breaking down siloes

Industrial use of APM has typically focussed on asset reliability and maintenance cost reductions for physical assets. The end goal is improving availability and reliability during an asset’s economic lifecycle. These solutions are superior to run-to-fail or other reactive asset management approaches. The downside, however, is that implementation is slow, complex and very expensive.

Keeping a system running, up-to-date and secure requires a significant investment in both IT and operational expertise. Moreover, traditional APM solutions have large capability gaps and do not provide the actionable information required to help plant managers make confident decisions about the best ways to run their assets to meet production schedules, or fully understand the financial results of those decisions.

While most plants run efficiently, there are numerous improvement opportunities, many of which rely on a fundamental change in approaches to IT and infrastructure. Today, too many are siloed, with plant data and systems often operating in isolation. This is a big problem – visibility of operations, delivery of production certainty and the ability to improve outcomes depend on the integration of systems. Decision support tools such as condition-based monitoring and predictive asset management with operational settings, process data, production information, and maintenance schedules need to be able to share information with each other. The creation of this kind of connected plant infrastructure is essential if APM is to deliver the quality of data required for self-healing plants to become a reality.