Revisiting Predictive Maintenance: Strengths of Predictive AI in Water Networks

What are the benefits of using predictive approaches in water maintenance? What can applying AI algorithms to water data help to achieve? What are the mistakes that can be avoided for predictive maintenance in water networks? Let's look at these.

Benefits of Predictive Maintenance

Predictive maintenance can be used to maximise system uptimes - this means the amount of time that the network, the area of the network or the specific piece of equipment is working, as opposed to its "downtime", when it is out of order or in repair. Depending on what the equipment is this can affect the whole network, perhaps even blocking it. Another benefit is "leveraging the service life" of each of these pieces of equipment. It's a "if it's not broke, don't fix it" kind of thing.

Think of it as like, not repairing something before it needs to be. 

Like imagine your phone.

Say someone told you you needed to replace it once every two years.

That would be a waste, right? It would be a waste if it was still working. Same with your fridge, or your car, or something...

Scheduling Jobs When Demand is Low

One last thing that predictive maintenance can help us to do, one of the benefits of AI in water distribution networks, is that when we know that a fault is coming and we can schedule the jobs then we can schedule the jobs when demand is low. This can reduce overall downtime in the water network, and outages.

Apply AI Algorithms to Data

Applying AI algorithms to data in digital twins enables us to schedule repair jobs for pieces of equipment before they fail but not so soon as to waste the pieces of equipment and to not use their full service lives.

Avoiding Mistakes

The biggest mistake that is often made in water infrastructure maintenance is fixing a set amount of time before a component is replaced, repaired or upgraded. Making this assumption can either cause a piece of equipment to

• Be upgraded too early, resulting in inefficiencies
• Be repaired too late, causing catastrophic failures.

By using AI to predict the health and lifespan of water network components, we can avoid fixing them too soon, but also predict faults before they happen, resulting in improved water distribution results.

This can show us how AI can be used further in predictive maintenance in water utilities.