scroll-up

8 January 2020
From predictive maintenance to predictive engineering

We normally see predictive techniques and tools applied to maintenance, monitoring in-service equipment to define when repairs should be scheduled, and avoid failure or downtime. The same ground concept – leveraging data to understand what happened and why, and what might happen next – is now being used to innovate engineering processes, thus improving the design and development of new products and related components.

Predictive engineering combines physics-based simulations with analytics, data mining, advanced statistical modeling and machine learning, providing companies with a wider range of information to be analyzed in the early stages of product creation. This is useful when the goal is a complex device integrating mechanical, electric and digital systems, so testing processes might be particularly complicated and time-consuming.

Thanks to new techniques, simulations can be completed at different levels with great accuracy, reducing time and effort to achieve an optimum result. The same occurs when a new material needs to be experimented, or the interactions between multiple devices are to be tested. Creating a common database environment allows engineering to fulfill large simulations and leverage Big Data to get key technical and performance metrics. Design steps can also be fast-tracked by running parallel simulations with varied parameters and comparing results before moving on with the following tasks.

What if predictive engineering could enter the product itself? We might have an on-board system detecting how a vehicle is approaching a curve and predicting it would not be able to hug the road under rain conditions, so alerting the driver to correct the manoeuvre. Another possible application comes from the possibility to get a warranty and support data into the play. Analytics may be used not only to recommend a component to be serviced before it breaks down but also to address the product design, so considering potential issues from the very early development phases. This might have an impact on product success by reducing post-sales maintenance costs, at the same time improving customer satisfaction and retention.

Predictive engineering requires a holistic and synergic view of the product lifecycle, bringing different data sources together and managing them through smarter methods and tools. As we saw for maintenance processes adopting a predictive logic, it requires important management and operational shift – as well as adequate skills and competences.