Pioneered by the process industries a decade ago, smart manufacturing is now being adopted by a much wider range of business sectors. Given the compelling benefits promised, it’s easy to see why. Extracting insight from the rich real-time data generated by production equipment offers the prospect of reduced plant downtime, greater manufacturing efficiency, and improvements in product quality that realize significant warranty cost savings. However, in the rush to implement, there is a real risk that enterprises are failing to convert ambitious intentions into a meaningful ROI.
Ironically, one of the most common pitfalls is a simple failure to identify, at the outset, exactly what that ROI will represent. Too many businesses are setting out to reap data, and only then trying to determine when and where the payback will materialize.
Another typical failing is to regard smart manufacturing as nothing more than aggregating sensor data on a desktop. In itself, that raw data is of little value. To reap any sort of return, companies need to adopt a much more considered, multi-stage approach. Specifically, it takes the mastering of capabilities in five distinct areas:
- Collecting data from a diverse ecosystem of connected equipment. Existing plant equipment may already be generating data. If not, sensors can typically be retrofitted.
- Preparing data by converting it into actionable formats. Consistency is vital for machine learning (ML) algorithms to be employed efficiently.
- Processing data by creating secondary measures. Bringing together multiple data streams often helps resolve otherwise unseen problems and inefficiencies.
- Monitoring data for anomalies. Real-time performance is key here; issues can be addressed before they become financial liabilities.
- Conducting analytics to correlate outcomes with anomalies. ML enables a relentless focus on the factors that impact efficiency and quality.
Crucially, the entire process should be shaped by domain expertise. Successful smart manufacturing programs are almost invariably driven by and from the shopfloor, not imposed by data and IT specialists. In particular, developing the algorithms from which insights are delivered requires detailed understanding of the physics at work in the manufacturing process. Moreover, giving ownership of the system to the people responsible for making smart manufacturing work will go a long way to fostering buy-in.
Ultimately, real-time analysis of manufacturing process data should be a game changer.
Instead of reacting to problems, production teams are empowered to predict and pre-empt them. Firefighting becomes a thing of the past. But that does not make smart manufacturing a simple, bolt-on solution to the myriad challenges of modern industry. Sustained payback will always require a thorough understanding of how raw data is transformed into actionable insights. Given the importance of a multi-stage approach, manufacturers also need to give careful consideration to their choice of system supply partner. Above all else, the ability to design and implement a solution that effectively integrates all five key steps will be fundamental to maximizing ROI.