Published

DETECT and save with prescriptive maintenance

Maintenance is essential. Excessive cost and downtime aren't. Released later this year, DISA's Monitizer® | DETECT prescriptive maintenance service will monitor every bit of sensor data from your DISA moulding machine in real time, immediately alert you to issues and add advice from DISA's expert engineers on the best action to take.

Foundries deploying this AI-driven service will be able to detect and correct any problems early while continuously optimizing equipment operation. Interested?

Smart foundries choose AI-driven prescriptive maintenance

When you can't see how vital internal components are performing, how do you keep sophisticated foundry machinery performing optimally at all times with minimum downtime - and at the lowest possible cost?

Bringing AI technology, huge data volumes and the Cloud together with human experience and expertise, provides simple answers to these technical challenges. Rather than struggling with complex analytics, foundry managers and service technicians instead receive timely alerts and advice that helps them spot and fix abnormal machine behaviour before it becomes critical and halts production.

Maintenance is evolving fast

In foundries and across industry in general, more accessible and affordable Industry 4.0 technologies are rapidly changing how we maintain our machines. Traditional preventive or scheduled maintenance is typically organized based on elapsed time or production volumes: after a set number of operating hours or units processed, technicians must inspect the machine and replace certain service components.

This is wasteful. Components will inevitably be replaced too early (or too late), the machine becomes unavailable for production and, worse, foundries will still experience unplanned downtime and machine damage where a part fails earlier than expected.

Condition-based maintenance is more effective, using KPIs based on sensor data to monitor the machine and schedule maintenance based on components’ actual condition. Essentially, the machine is “telling” you when it needs attention to prevent failure.

However, only critical wear parts or data relevant to standard maintenance tasks are normally monitored and the process is still reactive, with alerts usually indicating that a serious fault is occurring now or has already happened. Increasing sensitivity with lower alert thresholds in order to catch faults early often results in nuisance alarms. Most condition monitoring systems require significant IT investment and the in-house expertise to interpret the data correctly and avoid false positives. On top of that, it's only possible to manually interpret a small fraction of the data that's available.

For example, DISA's Remote Monitoring Service employs human experts to interpret machine operation and condition data and so provide timely advice. This works well - but even DISA's expert engineers can't manually analyse all the data available. With the many sensors now installed on DISA equipment, thousands of data points are generated for every machine cycle.

Moving to next-generation analytics

Modern IT enables the next step forward: predictive maintenance. By combining massive processing power, inexpensive storage and automated analytics like Artificial Intelligence (AI), engineers can analyse historic operating data to build a digital model that maps the complex relationships between many different parameters and operating conditions, and shows how they affect results like machine performance, component wear and so on.

Adding live data to this digital model turns it into a real-time tool, giving very early warning when a part is starting to fail or if the machine performs poorly. This helps minimize planned, and unplanned downtime while continuously ensuring stable performance.

But the predictive modelling process is complex, requiring much more data expertise and IT spending than condition monitoring. Even if an external supplier delivers the analytics as a service, interpreting the results and turning them into effective action still demands significant - and rare - in-house experience. With the high investment required, only the largest businesses can realistically take advantage.

Prescriptive maintenance overcomes that challenge by simply telling foundries when they have potential machine faults or poor operation, and what they can do to fix them. Because data collection, AI-driven analytics and human expertise are all part of the service, the foundry doesn't need extra IT or data experts.

With an easy-to-use prescriptive service, you get the same benefits as in-house predictive maintenance tools at much lower cost and business risk. Maintenance actions are not prompted by potentially damaging wear or serious component problems - as with condition-based maintenance - but by automated analysis of data from many machine sensors. Together, they provide very early warning of developing faults or sub-optimal operation.

So, foundries only take action to maintain a machine when it is needed, not before, and problems are usually fixed before they actually occur - preventing potential machine damage, downtime or higher scrap altogether. And as well as remedying problems early, foundries can also use the information to continuously improve machine operation and their process.

Monitizer®| DETECT, the first prescriptive maintenance tool for foundries, will be launched by DISA later this year.