How energy-based maintenance cut downtime and costs in tyre manufacturing

This article presents a case study involving heavy machinery, specifically rubber mixing machines (Banbury mixers). Rubber mixing is an energy-intensive process where large motors and hydraulic systems combine raw rubber with additives at high temperatures. 

These mixers are the backbone of tyre factories, and any unplanned downtime can be extremely costly. Additionally, they consume vast amounts of energy and are notorious for wasting energy when operating inefficiently. This made them an ideal candidate for testing the Energy-Based Maintenance (EBM) approach. By analysing their energy signals, inefficiencies and faults can be detected earlier than with traditional methods. 

The case study demonstrates how monitoring energy – in this instance, hydraulic power – combined with machine learning analysis, led to significant improvements in maintenance outcomes and efficiency. 

Implementing EBM

The team equipped the mixer with sensors to continuously record primary energy indicators, such as hydraulic pressure and flow, which together represent the machine’s power consumption within its hydraulic circuit. These readings effectively tracked the energy usage of the hydraulic system at any given moment. 

Data was collected via the machine’s SCADA system and dedicated sensors, providing a rich stream of information about each mixing cycle, including pressure levels, flow rates, and cycle times. 

It was not merely a matter of collecting energy data but of analysing it. Using machine learning (ML) algorithms, the team sifted through the data to identify patterns associated with normal and faulty operations. 

Given the numerous potential features that could be derived from the raw energy signals, Recursive Feature Elimination (RFE) was applied. This technique automatically selects the most informative features, helping to pinpoint which aspects of the hydraulic power signal – such as specific pressure spikes or flow fluctuations – were the best predictors of machine health issues. 

With these key features identified, ML models, including methods like Random Forest and neural networks, were trained to distinguish between healthy operation and various fault conditions. 

Findings

Detecting faults and predicting failures: Implementing EBM on the mixer delivered impressive results. The ML models learned to detect subtle changes in the hydraulic power signature that indicated developing problems. 

The system identified these anomalies as soon as they emerged, enabling maintenance technicians to intervene before major jams or failures occurred. This marked a significant improvement in fault detection, as issues that might previously have gone unnoticed until a stoppage were now flagged early by changes in energy behaviour. 

The case study demonstrated enhanced fault diagnosis capabilities through the EBM approach. 

Efficiency, savings and sustainability

The success of any maintenance initiative is measured by tangible benefits. In the rubber mixer case, adopting an energy-based approach delivered clear, quantifiable improvements. Below are the key outcomes and their relevance for manufacturing SMEs: 

• Improved fault detection: EBM identified faults that traditional monitoring methods might miss. For example, it detected early signs of a hydraulic leak when the pump began consuming slightly more energy to maintain pressure. It also flagged jamming events in the mixer’s operation by recognising irregular power usage during what should have been a steady process step. 
  By comparing energy profiles, the system differentiated normal variations from abnormal ones with high accuracy. This resulted in fewer hidden issues, as almost every anomaly in the machine’s performance left an energy trace that EBM could detect. Maintenance teams gained richer insights into machine conditions at all times. 
• Early failure prediction: Beyond detecting issues, EBM forecasted failures before they caused downtime. For instance, it predicted pump degradation several operating hours in advance by analysing energy consumption patterns. 
  This early prediction proved invaluable, transforming unplanned downtime into planned downtime. It prevented secondary damage by servicing components before catastrophic failure and reduced the ripple effects on production schedules. Essentially, the rubber mixer received a “health report” projecting future issues, enabling management to act strategically. 
• Reduction in energy waste: A standout benefit was the direct drop in energy consumption after implementing EBM. By identifying and addressing energy-inefficient conditions – even minor ones like a mis-tuned valve or a clogged filter – the machine used less power for the same output. 
  Over the course of the study, the rubber mixing process achieved a 3% to 7% reduction in energy use compared to previous operations. 
• Sustainability impact (CO₂ emissions cut): Energy savings translated into environmental benefits. Using less energy reduced greenhouse gas emissions for each ton of rubber produced, particularly when electricity was sourced from fossil fuels. 
  In this case, the 3–7% energy reduction directly equated to a 3–7% cut in the carbon footprint of the mixing process. 
• Strategic and financial value: For SMEs, the most compelling outcome was the boost to the bottom line and improved risk management.

Fewer unexpected breakdowns meant less unplanned downtime, enhancing overall production output and on-time delivery for customer orders. Energy savings lowered operating costs, turning maintenance into an investment rather than an expense. By converting energy data into monetary terms, management could see the euros saved on energy bills and the avoidance of costly repairs or scrapped batches resulting from late detections. 

Conclusion

This case study highlights that for manufacturing SMEs, EBM is not an overhyped concept but a viable strategy that can be implemented using existing sensors and data platforms, such as SCADA systems, alongside data science support. 

The result is a maintenance operation that is more predictive, efficient, and aligned with business goals.