arrow-right angle-down-sx angle-down-dx menu-hamburger search
test
test
test
test
test
test
test
test
test
test
test
test

Using Big Data to make Prysmian’s factories even more efficient

The Digital Innovation Lab is working with teams at two Italian plants on a Predictive Quality project using advanced statistical algorithms and machine learning to diagnose quality issues before they happen


Process engineers at the FOS optical fibre plant near Battipaglia and at Arco Felice are discovering how to squeeze more waste out of production by eliminating the guesswork when then diagnose glitches linked to the four “Ms”: men, machines, method and material. Supported by partner MoxOff, a universityspin-off, Prysmian teams are crunching data produced at the two factories to achieve three goals.

Meeting business needs with Artificial Intelligence solutions

Meeting business needs with Artificial Intelligence solutions
Two of Prysmian Group’s flagship factories are discovering how to become even more efficient by analyzing their production data to improve manufacturing processes. This Big Data project is being carried out in Italy at the FOS optical fibres plant in Battipaglia and the submarine cables factory at Arco Felice using new enhanced tools based on advanced statistical algorithms and machine learning, in order to prevent quality issues, reduce costs and improve operations.
The Prysmian team working on this Predictive Quality project includes people from Digital Innovation Lab, HQ Quality, FOS Quality, FOS R&D, and Arco Felice Process Engineers, in partnership with MoxOff, a spin-off created at Politecnico di Milano to apply mathematics to industrial processes.  The first round of analysis performed over the last month went well, and the project has now entered a second phase at FOS, where the team is implementing a customized solution to support the plant in reaching a challenging business objective.
 
Predictive Quality” solutions are used to help manufacturers gain insights into the quality aspects of their production processes by crunching their data and applying predictive algorithms in order to reduce losses incurred due to quality issues, and to recommend corrective actions.  Even at a high-tech factory, production problems can be traced to four main areas: men, machines, method, or material. In other words, human error, a machine glitch, a faulty operating procedure, or inadequate raw materials. Using Big Data analysis can help manufacturers get ahead of the curve, predicting these problems rather than reacting to them.
 

 

“We believe in a step-by-step approach, in which the knowledge and understanding of the problem is key,” says Carlotta Dainese, head of the Digital Innovation Lab. “We try to respond to the needs of plant managers and their day to day issues by including process experts from the start.”

The first step, she said, was to collect data and verify its quality. Then the team carried out “data cleaning” and data fusion to get a uniform foundation for analysis. A visualization interface was created.  

The project has three main goals: to support process control by giving engineers a hand in detecting anomalies; reduce the time needed to take decisions or to implement fixes because of new insight on root cause analysis; and optimize operations management by making the process more efficient through a system dashboard. The result should be a reduction in scrap, energy and resources.

 

“The goal is to produce a layer of Artificial Intelligence and obtain software that will be integrated in our production systems.
The project has a strong link with the plant’s business challenge. We don’t want to just enhance the value of data on its own. We want to meet business needs.”

Array

Carlotta Dainese

head of the Digital Innovation Lab

FOS, or Fibre Ottiche Sud, produces about 50% of all the optical fibre made in Europe by Prysmian Group, and Prysmian is Europe’s largest producer of optical fibre.

“Reaching important results at FOS means to transform this initiative into a strategic program for Prysmian,” says Prysmian Group Chief Strategy Officer Fabio Romeo. “Thanks to this experience we will identify a portfolio of models and tools, the needed competences and the main operational requirements to replicate this solution in other plants.”
“This Digital Innovation Lab initiative will accelerate our journey towards a new data-driven culture: the Predictive Quality solution can complement our strong business expertise with data-based knowledge to confirm our business assumptions, filter historical bias and discover some ‘hidden’ insights,” says Prysmian Group Chief Digital Officer Stefano Brandinali.

The Predictive Quality project represents also a cultural change for Prysmian, as well as a huge step forward to preventing problems in factories, says Group Quality Director Valentina Ghinaglia.
“Developing new tools based on data analysis is just a first step, if we are then able to integrate in our models the know-how of our engineers and operators, this will give us an unprecedented level of control on our processes,” she says. “The involvement of the local factory team is fundamental from the very beginning. The outcome will support the daily activities, allowing our people to take fast decisions, minimizing the risk of unforeseen events and lastly assuring the best quality of our products.”