Target audience:

Decision makers (Program Managers, Directors, CEOs) of manufacturing & allied firms, management consultants, manufacturing consultants, business strategists, innovators, and curious people.

Reading time:

5-10 min.

Introduction

Predictive maintenance (PdM) automation is an advanced technology tool that can help industries manage equipment health, reduce downtime, and optimise operational costs. To understand this automation tool better, it is important to understand the fundamentals, technologies, implementation steps, and prevalent issues to launch a successful PdM automation project. The below discussion provides a detailed explanation on this subject.

What is Predictive Maintenance Automation?

Predictive maintenance is a strategic technological tool that uses real-time data, machine learning, and predictive analytics to forecast equipment failures before they occur.

In the case of a preventive maintenance system (PM), a fixed schedule is followed for replacement of spares and associated maintenance, i.e., a feedback loop of machine performance from the respective machinery is not involved.

Whereas a PdM system continuously monitors machinery and triggers a maintenance alarm only when necessary. The key advantage over PM is—automation in PdM leverages IoT sensors, AI, and cloud platforms to streamline data collection, analysis, and decision-making. In other words, PdM enables maintenance departments of companies to act swiftly and efficiently to purchase required spares in advance and to schedule machine maintenance that ensures highest uptime of the factory’s machinery.

Why Should You Choose Predictive Maintenance Automation?

Notable features of predictive maintenance automation:

1. Reduced Downtime: Early detection of issues helps to prevent unexpected breakdowns.
2. Lower Maintenance Costs: Maintenance is performed only when needed, thereby saving on emergency repairs and reduces the number of spares purchased.
3. Extended Equipment Lifespan: Optimise maintenance schedule to keep assets in peak condition during high load periods.
4. Improved Safety: Hazardous situations (due to a possible machinery malfunction) can be identified before they can cause an accident.
5. Enhanced Productivity: Fewer breakdowns result in uninterrupted operations, high worker productivity, and on-time delivery of products.

Key Technologies in Predictive Maintenance Automation

Modern PdM projects rely on a suite of advanced technologies:

Source: Zenith Software

How Should You Launch a PdM Automation Project?

A Step-by-Step Guide:

Firstly, you should assess the current maintenance practices. You can begin by evaluating your existing maintenance strategies (historic data), identify gaps, note the recurring issues, and areas where downtime or costs are the highest. This assessment can help you set a clear objective map for your PdM project.

Secondly, choose the appropriate technologies. That is, IoT sensors, AI-driven analytics, and cloud-based platforms can be used as they fit your operational needs. You may consider factors like compatibility with existing systems, scalability, and ease of integration, tech savviness (technology awareness) of your staff.

Thirdly, invest in data acquisition and purging tools—install sensors on critical equipment to collect data on temperature, vibration, pressure, and energy consumption. Then, cleanse the data by removing outliers, correcting errors, and to ensure consistency. It is important to note that high-quality data is essential for accurate predictions.

Fourthly, identify conditional indicators. For instance, you should be in a position to identify and distinguish between normal operation and fault conditions using time-based and frequency analyses. Some of the common indicators include abnormal vibration patterns, temperature spikes, and pressure fluctuations.

Fifthly, train the predictive models. The historical and real-time data can be used to train machine learning models. For example, algorithms such as regression, classification, and clustering help forecast failures and estimate the remaining useful life (RUL) of components.

Sixthly, integrate PdM with your existing systems for automated scheduling, reporting, and asset tracking. Integration is vital to ensure that your PdM solution works seamlessly with your existing Computerized Maintenance Management Systems (CMMS) or ERP software.

Seventhly, train your employees. Identify technology enthusiasts in your maintenance teams and educate them on the use of PdM tools and to interpret predictive insights. This training ensures that staff can respond effectively to alerts, optimise maintenance actions, and contribute to on-going improvements of the PdM tools.

Finally, monitor, optimise, and retrain the PdM tools and staff. The staff involved should continuously monitor equipment performance and refine predictive models as new data becomes available. If necessary, staff should be re-trained to effectively  use the PdM tools. Regular retraining improves the accuracy of PdM tools and helps it adapt to evolving operational conditions.

Common Challenges and Solutions

Source: Xempla

A Brief Outline to Initiate a Predictive Maintenance Project

Steps to Launch a Predictive Maintenance Project [IFM]

1. Is PdM a good fit for your factory? Analyse historic maintenance data to find a clear answer (i.e., Does the ROI justify your investment?).
2. Select critical machinery (strategic assets) for a pilot project.
3. Determine the metrics, sensors required, and tools to be used.
4. Collect the data to be and plan for the required analytics.
5. Optimise the PdM tool as necessary to obtain accurate results.
6. Now, you can scale such PdM projects across your organisation.

Essential Predictive Maintenance Technologies [IFM]

Video Lesson: How to Use Machine Learning for Predictive Maintenance?

Video Lesson: How to Use Machine Learning for Predictive Maintenance

Case Study: Predictive maintenance at BMW Group Plant Regensburg

A Vehicle Attached to a Mobile Load Carrier or Skid System at an Assembly Line [BMW]

The Challenge:

1. Automobile assembly at BMW Group Plant Regensburg involves vehicles that are generally attached to mobile load carriers or skid systems.
2. These skid systems pass through the production halls on a chain. Any disruption or trigger of an alarm due to a technical fault in the high-tech conveyor systems can bring the assembly lines to a standstill.
3. On an average, the disruption to production due to unplanned maintenance results in a loss of more than 500 minutes of productive work time per year.

The Solution:

1. The company (BMW) implemented an AI based preventive maintenance system.
2. This predictive maintenance system monitors the conveyer technology during the entire vehicle assembly process.
3. The AI based system uses the existing data obtained from the installed components and conveyor element; consequently, it does not require any additional sensors or hardware.
4. Next, the data collected from the various sensors is transmitted to the BMW Group’s own predictive maintenance cloud platform.
5. Now, the AI algorithm continuously monitors for irregularities, such as power consumption changes, conveyor issues, or unclear barcodes that may cause malfunctions.
6. If an anomaly is found, the 24×7 maintenance control centre receives a warning message, which is assigned to the maintenance technician on duty.

The Results:

1. The AI-based predictive maintenance system is a cost-effective solution—no additional sensors, only storage costs and computing power are required.
2. Machine-learning models were developed in-house; the historic data is used to develop heatmaps for numerous abnormalities that could result in equipment failures.
3. Therefore, it helps the maintenance department identify different fault patterns in various components and respond to them through a systematic approach.
4. After this project was implemented, at least 500 minutes of downtime per year was avoided in vehicle assembly alone at the BMW Group Plant Regensburg—where a vehicle rolls off the assembly line roughly every minute.

Conclusion

To summarise, predictive maintenance automation is a powerful tool that can be used to reduce downtime, cut manufacturing costs (eliminate non-value-added tasks), and improve asset reliability. By following a structured approach and effectively using advanced technologies like AI and ML, even novices can successfully implement PdM projects and attain substantial operational improvements.

References

1. https://zenithsoftware.ca/a-beginners-guide-to-predictive-maintenance-technologies/
2. https://www.einfochips.com/blog/a-5-minute-guide-to-utilizing-predictive-maintenance-in-industrial-automation/
3. https://www.xempla.io/forever-forward/articles/predictive-maintenance-beginners-guide
4. https://www.automate.org/ai/industry-insights/getting-started-with-ai-based-predictive-maintenance
5. https://intellisoft.io/predictive-maintenance-software-development-a-complete-guide/
6. https://www.ifm.com/us/en/us/predictive-maintenance/implementing-predictive-maintenance-a-strategic-guide?srsltid=AfmBOoqqbDif3lwo-OQB219Xr_WO4rgwIFyIeSww02CXnQSiyks-I-qz
7. https://www.automationworld.com/process/plant-maintenance/article/21615374/a-step-by-step-guide-to-predictive-maintenance
8. https://www.youtube.com/watch?v=zXcp2HvpJLE
9. https://fiixsoftware.com/maintenance-strategies/predictive-maintenance/
10. https://www.press.bmwgroup.com/global/article/detail/T0438145EN/smart-maintenance-using-artificial-intelligence?language=en