Predictive maintenance and condition based monitoring (CBM) are two maintenance techniques rapidly gaining traction in packaging. Whilst these focus on keeping production lines running and minimising downtime, there is an advanced option, Prescriptive Analytics.
Prescriptive analytics uses a combination of machine learning, algorithms and business rules to simulate a multitude of “what if” scenarios. The outcome results in moving from what might happen (predictive) to what to do in order to achieve a specific outcome. This ultimately leads to reduced waste, improved efficiency, higher quality and increased RoI.
Examples of where prescriptive analytics can help in packaging are covered below. But in reality it’s all about how data is used – not just collected – to support pro-active decision making. This means fewer disruptions, minimised downtime and production processes running at maximal efficiency.
Key benefits in packaging
- Dynamic production scheduling: with packaging lines generally facing high-mix, low volume demands, prescriptive tools can automatically reconfigure schedules in real-time. For example, if a rush order is received the most efficient and cost-effective transition between products can be determined.
- Inventory optimisation: ensures the right quantities of products are stocked freeing up capital to be used elsewhere. This is vital if there are short production runs with frequent change-overs, allowing flexibility and adaptability to customer needs.
- Quality control: instead of waiting to catch a defective product at the end of a run, prescriptive systems can detect micro-fluctuations. For example, pressure or temperature, prescribing machine adjustments to keep products – such as liquid-fill in bottles – within specification.
- Maintenance scheduling: whenever there is a breakdown, every lost minute is costly. Condition based monitoring monitors asset health whilst predictive maintenance identifies when an asset might fail so preventative action can be taken. However, it’s still important to minimise disruption to production. Prescriptive analytics can recommend the best time to perform any maintenance without disrupting production targets.
So how does it work?
- Data input: data is gathered from various sources, such as IoT sensors on machines and the factory floor, ERP (Enterprise Resource Planning) systems and warehouse Management Systems.
- Modelling: machine learning tools, AI and optimisation algorithms are used to simulate different scenarios.
- Actionable output – turning predictions into decisions: specific recommendations can then be made e.g. when to schedule maintenance for a PacDrive M SM motor on Line 1.
All this may sound overly challenging when many packaging companies are only just embracing condition monitoring, let alone predictive maintenance. Adoption of even these techniques requires time and planning which can be severely tested when unpredictable day to day issues demand attention.
However, any technique or process which can minimise unplanned downtime and optimise processes – on the factory floor, inventory management and even energy usage – is something worthy of serious consideration. Packaging is a highly competitive sector, often operating on low margins. On the one hand this could hamper investment, but on the other potentially motivate forward-looking companies to seriously look at ways they can improve their competitive edge