Companies are looking to digital transformations strategies for predictive maintenance tools in the manufacturing industry. Here’s why.
by Sean Molloy
It’s common knowledge that in order to extend the lifetime of a machine or system and to keep them functioning at optimal levels, they must be maintained. This is just common sense and we have all expended real dollars to fix broken items that were not well maintained. But what’s not common sense is how we decide when a failure is eminent, and what maintenance must be immediately executed. Companies are looking to digital transformation strategies for predictive maintenance tools in the manufacturing industry to answer these questions.
Original manufacturers of equipment provide manuals and operating guidelines that often reflect predictive maintenance schedules when the best operating conditions exist. But what about real-world conditions where many more uncontrollable factors are present? Do manufacturer’s recommended predictive maintenance schedules really apply, are they more like guidelines, or are they of little value?
In order to deal with maintenance under real-world conditions and prevent catastrophic breakdowns, organizations need much more information along with new tools. The information requirements include the standard measurements the manufacturer provides and extends to data that is not readily available. The tools requirements include standard machine sensors and dashboards as well as additional sensors and visualization tools. However, most organizations are not holistically addressing how they keep their mission-critical equipment and processes functioning. They rely on siloed techniques and tools.
When we ask most of our customers to identify the critical equipment and processes that have the largest cost and loss of capacity if they fail, they all easily identify these pain points. They tell how it impacts the plant and disrupts their supply chains and customers. It is important to note that many of the critical items identified are fraught with the dangers that can possibly lead to larger catastrophes including loss of life.
When asked how much it would save the organization if they could prevent the failure, their answers often fall in the hundreds of thousands or millions of dollars per day/week. It is surprising how fast they can quote a real dollar amount and the associated business impact. The information is well known, and the return on investment (ROI) is easily agreed upon by the business leaders.
When asked if they know what the signs are that that tell them something is going to go wrong, most tell me that they do know the key factors but don’t have a way to measure or monitor them in a meaningful way. This is because not all key factors have sensors on them and many of the key factors lie upstream or downstream of the critical equipment.
Let’s pause here. Notice that each paragraph starts with when we ask. Also notice that the business readily knows the answers. This is significant because we don’t need to do extensive studies or work tirelessly on a business case. If the big-ticket items that everyone is very aware of—and can agree is a priority with significant financial impact—are addressed first, there is typically universal agreement to move forward with those projects. In other words, companies know where the big risks are and what the costs are if there is a failure. Manufacturing companies are starting their shop floor digital transformation journey in this area as these projects are easiest to secure agreement on purpose, scope, deliverables, and ROI.
The next set of questions are geared to the equipment and process experts. We ask them to tell us the factors that can alert us to failure before it occurs. This includes factors within equipment, processes, and environmental factors. Once these are identified we discuss what type of sensors are present and what type of sensors need to be added as well as where they should be placed. This is the simplified way of saying what Industrial Internet of Things (IIOT) should be included. Very often we find that original equipment manufacturers have added sensors to their equipment, but they often don’t include sensors where the customers need them. So, adding sensors to “smart” equipment is something to be expected.
Once we understand the parameters that need to be monitored and the data that we collect concerning those parameters, we decide if the data will be streamed and collected in a historian (cache)—allowing future analysis—or if the data will be live streamed and discarded. All of the data will be used by the models that are set up in predictive maintenance tools to monitor the process and provide alerts of possible failures.
Predictive maintenance models will be created and validated during one of the sprints in the project. These projects lend themselves to the agile implementation methodology. Each sprint focuses on a specific set of functional and technical deliverables. This methodology is great for managing the development and keeping the project focused on the timeline and deliverables.
The last concern is how users interact with the new intelligent enterprise solution. Dashboards for management are always beneficial and should be included in the project. However, dashboards are not a great tool for the maintenance technicians in the field. To bring it all together for the technician we need to introduce “mixed reality.” Tools such as HoloLens will bring the required information into a live view of the equipment and processes while they are running. The sensor data with operating performance and the predictive health models are all dynamically displayed as the technician looks at each part of the process. If any adjustment needs to be made, the HoloLens highlights those areas and can walk the technician through the repair or maintenance process. Additionally, a group of process engineers can remotely see what the field technician is looking at and provide real-time assistance.
With a proliferation of IoT devices that allow us to monitor just about any aspect of a manufacturing process and powerful predictive maintenance tools, most businesses can expect a return on investment in less than 1 – 2 years. It’s never been easier to get started.