Making plastics machinery more efficient with Predictive AI
Milacron’s partnership with ei3 is a true success story in innovative IIoT solutions for injection molding machinery. Here’s how M-Powered was developed to provide significant maintenance and repair cost reductions, uptime increases, and massive energy savings.
At a presentation delivered by Milacron at ei3’s ConnectedAI Summit, the history, purpose and workings of M-Powered were shared with industry insiders and experts.
The partnership between Milacron and ei3 started in 2014. Initially, it focused on connecting Milacron’s fleets of machines to provide secure remote services and basic IIoT capabilities (such as OEE calculations and downtime tracking) to their industry-leading injection-molding machines. However, the partnership quickly expanded to include ei3’s Data Science capabilities to put additional tools into the hands of customers that would provide actionable insights to help maintain their fleet. The analytic tools developed as a result of this decision were wrapped into a suite and released as ‘M-Powered’ in 2018.
M-Powered today represents the state-of-the-art in industrial IoT, providing a rich set of applications to track machine conditions, provide operational advice to operators and predictive information to service technicians, and a platform to efficiently and digitally manage and optimize all parts of the machine operation, using the power and strength of data science and practical AI. The suite works along three general processes: Transitioning machine data into the cloud, using analytics engines to turn data into insights, and providing real-time insights through alerts, desktop applications, and mobile applications.
A critical service highlighted in the presentation was ei3’s Lifecycle application, designed to track the deterioration of specific components of a machine to reduce costs and unplanned downtime.
Particularly notable is the use of the application for the maintenance of heater bands, a crucial component of an injection molding machine that heats the plastic granulate inside the barrel to a liquid state. The traditional maintenance method for heater bands is either based on a time schedule or operator discretion. Often, operators faced with minor alerts and error messages overlook or ‘clear’ them in order to prioritize uptime. Unfortunately, this leads to inefficient operation and eventually the component’s failure, which leads to increased costs and unplanned downtime.
Ironically, the difficulty of preventative maintenance for heater bands is a result of the machine’s automatic compensation for these failures: Faced with reduced heating power, the machines will cycle more energy into neighboring bands to compensate. The problem with this approach, aside from the danger of prompting more heater band failures, is that overworking heater bands result in a significant increase in energy costs, which dramatically outweigh the price of a replacement heater band.
Using predictive analytics and measuring the deterioration of not only heater bands but the behaviors of the overall process, manufacturers can be warned of the need to replace a heater band – before the component’s deterioration causes excessive energy costs that outweigh the cost of the replacement part.
Another prominent example of ei3 and Milacron’s application of predictive analytics and custom solutions is seen in the way M-Powered is capable of monitoring and tracking the degradation of feed screws. The degradation of a feed screw differs from that of a heater band. While heater bands will eventually fail entirely, the degradation of a feed screw leads to incremental effects on the quality of production.
This difference requires a new approach to monitoring in which an analytic method maps the root causes of degradation and creates a specific time of replacement based on a client’s particular needs in regard to quality. For example, by measuring the impact of the degradation of a screw on the quality of output, M-Powered could tell a manufacturer in the medical industry to replace feed screws once they’ve degraded to below 90% efficiency rates, while a manufacturer of home goods may replace them at 60% efficiency. This specific process is crucial for plastics suppliers, where plastic waste is a massive cost and where maintenance schedules directly correlated with quality management is a great benefit.
M-Powered’s Lifecycle app does this by detecting and quantifying the steady degradation of machine parts using system-wide measurements.
M-Powered delivers results in the real world. For example, Orbis, a large client of Milacron machines, uses M-Powered to predict and proactively replace machine components – and subsequently save 1000 kWh of energy on each machine, the equivalent of what an average American household uses per month. But not only that: The improvements at ORBIS also resulted in record low downtimes and a tighter, more exact maintenance schedule than before while simultaneously saving on inventory costs that accrue with stocking.
Predictive Services are just one facet of M-Powered’s wide array of IIoT services that help move plastics manufacturing into the future.