AI is undoubtedly already changing the way industrial organisations think about maintenance. And rightly so, as it packs quite the punch; analysing equipment data, identifying patterns and detecting potential failures before they occur. 

But, as with most things in life, the sum of the parts is what ultimately leads to success.  For AI to come to its fullest fruition, infrastructure, data and people must also fall into place. This holds particularly true for South Africa where many industrial facilities still comprise older equipment and communication systems.

“South Africa has a vast legacy of industrial installations, and the challenge lies in integrating these with modern technology,” says Paul Steyn, Field Services Projects Manager for SSA at Schneider Electric. 

“We are still heavily dependent on human intervention to operate equipment correctly. AI cannot simply be added to an industrial environment and expected to deliver immediate results.”

Three layers for optimised AI 

According to Steyn, effective AI implementation requires three critical layers: intelligent equipment, a robust communication network and a common operating system capable of converting large volumes of data into information that people can actually use.

“In a large plant, thousands of data points are constantly being sent to a central system. The challenge is deciding how to use that data. How do you translate it into information that makes sense for the operator on the floor, while also presenting it in a way that a CEO or CFO can connect to the financial performance of operations,” notes Steyn.

New industrial facilities can be designed around modern Ethernet, fibre and other communication technologies. Older facilities, however, may have accumulated multiple communication topologies and network infrastructures over decades.

“The first step is to properly assess the communication network infrastructure on site. From there, decide which elements you want monitored through the AI interface and predictive maintenance systems. Finally, make sure you have a robust, end‑to‑end communication backbone that connects every point reliably,” explains Steyn.

“The reality is that those companies who want to integrate AI and intelligent equipment need a strong on‑site communication backbone which often means upgrading or migrating to a modern topology that can support AI‑enabled systems. Schneider Electric can help facilitate this assessment to ensure the infrastructure is ready for next‑generation technology.”

Unlocking the early warning system

Where the foundations are in place, predictive maintenance can deliver tangible operational benefits to industrial operations.

For example, traditional planned maintenance typically works according to fixed intervals, with technicians conducting scheduled inspections and servicing. While undoubtedly necessary, these interventions can require planned downtime, even when equipment may not actually require attention at that point.

“An AI predictive maintenance system can flag potential failures in electrical equipment before they happen. This allows you to service equipment when needed, rather than on fixed intervals,” says Steyn.

However, it is also important to note that whilst AI can identify potential failure it cannot independently resolve the problem – human intervention is still paramount. 

“If an electrical motor shows signs of failure and AI predicts it will break down soon, human judgement is still needed to decide the best time to shut the plant for repair or replacement,” he explains.

“Predictive maintenance doesn’t eliminate downtime; it shifts it from sudden breakdowns to planned intervention, providing you with a window to plan more effectively. It’s an early warning system; not a guarantee against downtime.”

Building a confident and skilled workforce

Understandably, there is a concern amongst people working in industrial environments that automation and intelligent systems could eventually replace their roles.

Steyn however argues that the more realistic future is one in which skills evolve alongside the technology. Rather than removing people from the equation, organisations need to consider how their existing workforce can be equipped to work alongside increasingly intelligent systems.

“We’re seeing larger organisations with extensive installations and big workforces prioritising personnel training in order to not only support systems but also gain the most from AI integration. The goal is to better support their systems while effectively integrating AI,” says Steyn.

“In the end, industrial workers will still need to understand equipment, maintenance and safety, but increasingly they will also need to understand how software, data and AI interact with those physical systems,” concludes Steyn.