Get the latest updates on the Coronavirus impact on engineers.Click Here
Asset Management

Focus on asset health

Consider digital asset management systems as the heart of your Industry 4.0 transformation to ensure the health and competitiveness of your future business
By Andrew Kowalik March 12, 2020
Courtesy: Nokia

When companies can accurately pinpoint their high-risk assets, they can prioritize maintenance and repair work, justify and allocate their capital spending and increase their reliability and regulatory compliance. Unfortunately, judging asset health is as much art as it is science. Manufacturers’ maintenance schedules may do a good job of covering their liabilities, but they will empty your coffers. While the best machine operators have an intuitive sense about what it takes to keep things running smoothly, with increasing automation and workforce turnover that can’t be relied on the way it once could.

Automation aids asset maintenance

Fortunately, other aspects of industrial automation are making it easier to keep asset health in focus. Digital sensors, digital twins and software analytics combine to bring science into asset maintenance. In the process, they are revolutionizing how companies not only manage their assets, but also how they operate their businesses.

Traditionally, asset maintenance is accomplished by one of two methods: time-based or condition based. An example of time-based maintenance is the manufacturer’s recommended maintenance schedule for the equipment. Condition-based maintenance monitors the equipment, applies analytics and uses the insights generated to determine the optimal maintenance schedule. It is often the better approach of the two given the complexity of assets and operations today, the increasing digitalization and automation of industrial processes and the need for a reliable and flexible manufacturing environment, which calls for zero-downtime operations.

Digital twin for predictive maintenance

Industry 4.0 is changing asset management by introducing digital technologies that can monitor assets and optimize performance in real time. Machine and sensor data along with analytics can be used to create a virtual model or digital twin of the machine. The digital twin can then be used to assess the machine’s performance to more precisely pinpoint when it is likely to fail, commonly referred to as predictive maintenance.

There are many kinds of digital twins or virtual models depending on the problem being solved. In the case of predictive asset maintenance, the key is to compare actual performance data to historical performance data. With machine learning, the algorithms and the model become more precise based on actual experience. This added precision helps to schedule maintenance, repairs and replacements with greater accuracy, avoids downtime and helps with workforce management. It also can reduce the number of spares and the cost of ensuring redundancy.

The best analytics applications not only provide more precise estimates of when machines will fail, they will keep track of all your assets and focus your attention on where the greatest risks lie. This often-complex calculation for asset managers is trivial for an analytics application, once you’ve fed it the data it needs. The application keeps track not only of the risk of failure, but the time-to-repair and cost-to-repair and the likely cost of failure. It can then red flag issues posing the highest risk to the business.

However, this is only the beginning of an entirely different way of managing assets. Digital twins or virtual models don’t have to exist in isolation. It also is possible to connect digital twins and create a picture of your entire operation (see Figure 1).

Figure 1: Creating a digital twin of your entire operation by aggregating individual digital twins. Courtesy: Nokia

Figure 1: Creating a digital twin of your entire operation by aggregating individual digital twins. Courtesy: Nokia

The aggregation of twins creates system-level, factory-level and business-level views. This introduces the possibility of workflow optimization, where analytics programs can suggest areas for workflow improvement and redesigns that have factory-wide and business-wide impacts. This is where risk assessment models become valuable as the analytics programs can identify and measure the spillover effects of malfunctions and failures. This data can then be mapped to parameters such as cost, quality and even customer satisfaction.

There is never too much data for a software analytics program. For a human asset manager, the addition of all this extra data would quickly reach a point of diminishing return. An analytics program using machine learning and artificial intelligence has almost infinitely more capacity to sift through data, filtering, sorting and identifying action-impacting correlations (see Figure 2).

Figure 2: An analytics program using machine learning and artificial intelligence has almost infinitely more capacity to sift through data, filtering, sorting and identifying action-impacting correlations. Courtesy: Nokia

Figure 2: An analytics program using machine learning and artificial intelligence has almost infinitely more capacity to sift through data, filtering, sorting and identifying action-impacting correlations. Courtesy: Nokia

Monitoring and controlling asset health is a good beginning. But as you evolve your operations using Industry 4.0 technologies, it is only the start of a much more ambitious, long-term digital transformation of your business. It is important in thinking through your digital asset management systems that you think of them as part of a platform for your entire business with a single, powerful analytics engine at the core. It will be the heart of your Industry 4.0 transformation and ensure the health and competitiveness of your future business.


Andrew Kowalik
Author Bio: Andrew Kowalik is strategy head of industrial markets in the Enterprise Business Group at Nokia.