Podcast: AI-supported predictive maintenance: Preventing downtime instead of managing it

Listen to the "BULK TALK" podcast directly:

In the podcast "BULK TALK – the podcast for an innovative bulk solids and recycling industry," our managing director Sebastian Pohl spoke with Julika Hecht, senior marketing manager at SOLIDS & RECYCLING-TECHNIK Dortmund, about the future of maintenance.

The focus: AI-supported predictive maintenance in the bulk materials industry – and the question of how unplanned downtime can be systematically avoided.

For operators of complex packaging and filling plants, one thing is clear: it is not maintenance that causes the highest costs, but unplanned downtime. This is exactly where a data-based approach comes in.

In the interview, Sebastian Pohl describes the development of maintenance as an evolutionary process:

• Reactive maintenance: Repairs are only carried out in the event of a malfunction. This results in production interruptions, weekend work, and high follow-up costs.
• Preventive maintenance: Maintenance is carried out at fixed intervals, regardless of the actual condition of the components. This reduces risks, but sometimes leads to unnecessary interventions.
• Predictive Maintenance: Machine data, sensors, and AI models enable condition-based maintenance. Interventions are carried out when data indicates imminent wear or a malfunction.

A key topic of the podcast: AI only works on the basis of valid data. Our systems are now fully networked. Among other things, we analyze:

• Operating cycles and load profiles
• Motor and robot data
• Sensor values and process parameters
• Filling behavior and performance indicators
• Fault histories

By systematically collecting and evaluating this data, we can identify patterns that indicate impending wear or efficiency losses.

In this context, predictive maintenance does not mean "black box," but rather structured data analysis with a clear objective: to increase plant availability and reduce operating costs.

In the podcast, Sebastian Pohl introduces our VeloXpert Service.

Together with our partner EI3 (USA), we are developing a secure data connection platform that analyzes machine data and makes it available for optimization purposes.

The platform enables:

• Secure collection and analysis of machine data
• Transparency regarding OEE and plant performance
• Derivation of specific maintenance and replacement times
• Remote service and digital support

We already rely on complementary solutions such as:

• Remote support with a high problem-solving rate
• OEE dashboards for performance analysis
• Digital twins for identifying spare parts
• Velo-Guard for wear monitoring of shut-off valves
• Bag test bench for optimizing speed and accuracy in valve bag systems

The goal is clear: Maximum availability with predictable interventions.

A key consideration for technical decision-makers is return on investment. The podcast makes it clear that:

• A structured preventive maintenance strategy usually pays for itself within the first year.
• Investments in predictive maintenance typically pay for themselves in less than 18 months.

Unplanned downtime causes production losses, express spare parts procurement, personnel costs, and consequential damage. In contrast, there are predictable maintenance costs and digital services with clearly calculable expenses.

Medium-sized companies with a high degree of plant dependency in particular benefit from this predictability.

Sebastian Pohl emphasizes in the interview that AI is not a short-term trend. Companies must integrate AI structurally into their processes.

In addition to data-based maintenance, the following topics are becoming particularly important:

• Machine learning-supported plant optimization
• Remote service concepts
• Digital twins
• Virtual reality-supported maintenance support

Technological developments will enable significant efficiency gains over the next three to five years.