Florida-based design and manufacturing solution provider Jabil has been implementing solutions from Microsoft to create digital and predictive factories which use machine learning, predictive analytics and the cloud to learn ahead of time when a piece of equipment might fail.
“For the first time we have been able to predict, say a solder defect, or some type of component defect, on our printed circuit board assembly before it’s occurred,” says Clint Belinsky, Jabil vice president for global quality. “The fact that you can predict that and know why you predicted it is really impactful. Because the next step is to identify that as a problem and to provide specific instructions to eliminate it, so it never goes to a debug or a rework.”
Jabil is currently using Microsoft technology including Azure, Azure Machine Learning, Azure IoT Suite, the Cortana Intelligence Suite, Power BI, SQL Server, Windows 10 and Office 365 to create digital factories.
Microsoft highlighted that many manufacturers are using software to analyze real-time problems and defects in manufacturing in order to take corrective action. But Jabil wanted to create a solution that would identify errors or failures early in the process.
Jabil’s solution, which uses Microsoft Azure services, analyzes millions of data points from machines running dozens of steps through the manufacturing process, predicting failures earlier in the process.
The accuracy rate has been at least 80% in the prediction of machine processes that will slow down or fail, Belinsky says, contributing to a scrap and re-work savings of 17%, and an energy savings of 10%
“If a problem can be detected at step 2, instead of at step 12 or later, then you win, because it gives you the opportunity to stop and fix an issue, or stop a potential problem from happening,” Belinsky said.
“It may be taking out a tool and cleaning it … having to divert some raw material to a different location, or having to reject it,” he said. “Either way, you’ve avoided a problem, you’ve avoided downtime, you’ve avoided scrap. And that’s a very positive impact on the bottom line, and generates a lot of positive accolades from the customer.”
The platform is in use at Jabil’s manufacturing plants in Guadalajara, Mexico and Penang, Malaysia. Jabil said that it aims to roll out the solution to all facilities worldwide.
Jabil produces a broad range of products, from wearable technology, smartphones and medical devices to automotive, aerospace, defense, enterprise and infrastructure products.
