Unplanned downtime can be very costly and disruptive in multiple industries. This is particularly true for manufacturing, where unplanned downtime can cost a company as much as $260,000 an hour. Operators today may have poor visibility into the health of their manufacturing assets that can cause this unplanned downtime. These operators are focused on keeping the manufacturing line up and only performing maintenance either when an unplanned downtime events occur or on a time-based schedule. Implementing a predictive maintenance strategy can help improve the uptime, utilization, and performance of these assets [Fig 1].
Figure 1. Value of Predictive Maintenance on Asset Utilization
Predictive maintenance is a journey that is not completed in a single day. For some organizations, it may be difficult to even start this journey. For one, they may not have access to historical data such as sensor measurements on an asset, prior issues, and what corrective maintenance actions were taken. Even if this data has been captured, organizations may not have the right team members, skillsets, or tools to explore and analyze this data to identify patterns and occurrences to start formulating how one could predict issues before they occur. Finally, implementing a comprehensive predictive maintenance system from the beginning can be overwhelming both in terms of costs and scope that can be hard to justify investment of time and resources.
For an organization to start implementing a predictive maintenance strategy that faces the above challenges, they should approach this with 3 key facets: Start Small, Keep it Simple, and Learn Fast. Let’s dig into each of these areas:
- Start Small – for an organization, they may have multiple types of assets and potential failure modes they want to start to predict ahead time to minimize downtime. While they eventually will want to provide 100% coverage, they have to start somewhere. For this reason, I would start with one type of asset. As no prior historical data may exist, I would select the assets that have generally caused the most downtime and issues based on feedback from the operations team.
- Keep it Simple – organizations are concerned with the time and resources required for implementing predictive maintenance. This can be particularly challenging where personas such as data scientists or IT specialists may not be available to support this effort. With this, the goal should be to empower the experts – such as the operators and subject matter experts. Once the targeted asset is identified and data is captured, the goal should be to enable these experts to review and explore this data to help identify potential patterns and occurrences. This enablement should be done in easy-to-use tools and user interfaces with little to no coding or programming required.
- Learn Fast – organizations do not have infinite time and resources to explore and play around before they need to show a return on investment. Starting small and keeping it simple allow organizations to be very focused on investigating in one particular asset. This is one reason to select an asset type that may have more occurrences of an issue instead of the overall severity of an issue. Selecting an issue that occurs more often allows for faster identification of patterns of the issue, the ability to test and validate over multiple events that increases confidence, and demonstration of value when operationalizing the approach. In addition, this allows the organization to “fail fast” if there are no means to predict this issue early so that they can pivot and focus on different approaches. Overall, this allows the organization to demonstrate value and build momentum to tackle the next failure modes and asset types
As mentioned previously, predictive maintenance is a journey. By following these steps, an organization can quickly show value of predictive maintenance at a small scale with a minimum investment. Over time, organizations can continue to invest and expand by looking at different assets and failure modes. In addition, organizations can start to perform more remote and fleet management so that they can start to manage and optimize the utilization all of this from the asset level, facility, or across the entire organization! The hardest part is getting started.
Learn how your peers in manufacturing organizations are using a quick start solution for monitoring CNC machinery. At SAS Global Forum 2021, SAS in partnership with ADLINK and Intel shared a new way for organizations to start their predictive maintenance journey following the 3 facets I covered in this article. View the video here.
About SAS in IoT
SAS empowers organizations to create and sustain business value from diverse IoT data and initiatives, whether that data is at the edge, in the cloud, or anywhere in between. Our robust, scalable, and open edge-to-cloud analytics platform delivers deep expertise in advanced analytics – including AI, machine learning, deep learning, and streaming analytics – to help customers reduce risk and boost business performance. Learn more about our industry and technology solutions at www.sas.com/iotsolutions
Sanjeev Heda, Principal Industry Consultant, SAS IoT
