Experiment 12: Digital Brain For Predictive Maintenance in the Automotive Sector (DRIVEN)

Motivation of the Experiment

One of the main issues manufacturing businesses face is the operational impact of unplanned downtime.
While in the world of production, plant downtime is often planned, due to routine maintenance on equipment, upgrading or replacing machinery, implementing new processes or technologies, making repairs, etc.,  unplanned production stops have a huge impact in various terms. They can cause unforseen costs, security risks, and, on a higher level, the corporate image or the chances of losing business.

But even when downtimes are properly planned, no industrial plant is safe from an unexpected situation that can delay the plant's reopening. Any minimal unforeseen stoppage, of hours or days, is enough to generate losses in the millions.

The possibility of working in virtual environments, which manage to visualize scenarios approaching what the real situation would be, can give a great competitive advantage to the companies that implement it, in terms of: failure reduction, reduction in downtime, the increase in the economic useful life of the assets, and better security. DIGITbrain's experiment 12 aims to develop predictive mainteinance in PLC monitored factories by making use of advanced machine learning technologies. The final product is intended to predict when a system failure can occur and alert the factory correspondingly in order to reduce/avoid unnecessary production downtime.

Purpose of the Experiment

The experiment is aimed at avoiding efficiency losses in machinery, one of the main concerns in industrial processes. Under this premise, it analyzes, simulates and optimizes the entire product movement operation between a parts warehouse and the packaging process in an automotive component factory.
The purpose of this experiment is to be able to decrease risk of failure (ROF), increasing like this the overall equipment effectiveness (OEE) and so, increasing production. This will be achieved by making use of near real time predictive models that require HPC computing.
The experiment optimizes production processes, achieving a significant increase in OEE (Overal Equipment Effectiveness), an indicator that measures the effectiveness of industrial machinery, increasing the efficiency, productivity and profitability ratios of companies. The software required will be mostly but not uniquely based on python and it will be publicly available, along with LIS Data Solutions private tools combined.

Technical Impact

The experiment's technical impact is about achieving a "digital product brain", or a digital replica, which analyzes:

• Data throughout the life cycle of a production line or a machine
• The environments in which it operates.
• The systems that produce them.

With all this data, it can predict virtually everything that will happen in the physical world. The solution proposed in this project presents a technically challenging and potentially huge impact in industry. Fields such as the customization of parts, based on specific requirements, could significantly reduce their cost and will be more in line with customer requests. Likewise, having seen the consequences of the processes in advance, it will be possible to optimize all the areas involved. The challenge is to predict anomalous machinery performance simply by measured PLC data, and when successful, this project can extrapolate to many different industry scenarios that actually record such data during their processes.

Expected Economic Impact

It is difficult to foresee the final impact that the new technologies will have, related to the digital transformation in the new world paradigm. What we do know for sure is that technology is capable of making processes more efficient, and the intelligent use of data can give visibility to dark areas in companies. Bringing all this knowledge to light will undoubtedly achieve a more competitive and more prepared industry with a greater capacity to adapt to changes. The economic impact of the product we are developing is big for individual industries, since it can reduce the downtime of the machinery, and it can have a significant effect since it is intended to extrapolate to many modern industries. Any technology, that helps avoid uncertainty will undoubtedly be a competitive advantage for those that implement it.

"The implementation of the software developed by LIS data solution as part of the DIGITbrain-Driven experiment allows the online monitoring of the production line at SEINSA's factory. Furthermore, it estimates the OEE and the probability of an approaching machinery downtime up to a minute before the actual machinery failure."

Project Partners: