Five steps to reach smart predictive maintenance

Smart predictive maintenance accelerates the maintenance journey and has potential to increase machine availability and visibility across an entire asset network.

July 28, 2017

A blog post by Chris Coleman, specialist leader, Deloitte Consulting and Ed Deuel, specialist leader, Deloitte Consulting.

New techniques can improve plant throughput

Maintenance professionals today can face a number of issues, often including outsourcing, cost-cutting, scarcity of experienced labor and increasing complexity of equipment. Whatever the challenge, maintenance and reliability professionals share a common goal—to maximize machine availability. Yet traditional maintenance programs can only take you so far. In fact, machine failures go well beyond statistical time-based failure. Recent studies show that only 20 percent of machine failures are time-based, while the other 80 percent of failures occur either in the infant mortality startup phase or most often due to random or unknown failure.1 But truly, no failure is random, only that the root causes have gone unidentified. Modern maintenance techniques can help detect impending failures before they happen with typically more accuracy than time-based approaches. For manufacturers, exceptional asset maintenance can be a strategic differentiator in improving a plant’s throughput, efficiency, quality, and safety.

Accelerate the journey to world-class plant maintenance

Smart predictive maintenance (SPdM) may be able to accelerate the journey to world-class maintenance in five suggested steps. The technologies and principles discussed in this article are market-ready and have the potential to completely transform your maintenance operations and save millions of dollars for large companies at scale. However, we cannot overstate enough that technology alone is no substitute for a deep understanding of traditional maintenance theory and application. Figure 1, illustrates the six core maintenance pillars that every maintenance organization should consider developing (green circles) and the surrounding technology accelerators that can speed up results and multiply their impact (blue circles). Note that the pillars and technology accelerators all surround the nucleus, made up of a robust maintenance strategy and processes.

Figure 1: Maintenance pillars and technology accelerators

Without a well-defined strategy and mature processes, many may find that technology only complicates things. On the other hand, with a solid foundation and a clear, long-term vision, a technology solution can add exponential value to any company. And regardless of where you are in your maintenance journey, you can develop your end-state strategy and pilot a smart predictive maintenance solution in parallel.

Smart predictive maintenance defined

Smart predictive maintenance is a modern maintenance technique that leverages multiple technologies and maintenance approaches, including a powerful, advanced method, predictive maintenance (PdM). We have discussed the PdM approach and components in detail in "Predictive maintenance and the smart factory."

Definition of smart predictive maintenance (SPdM): The continuous monitoring and analysis of a network of assets, which enables the prediction and notification of potential failures, informs maintenance scheduling and spare parts planning, and automates some maintenance tasks.

Smart predictive maintenance goes beyond predictive maintenance in three ways:

  • SPdM monitors a network of assets, which are connected by the Internet of Things (IoT). IoT is not just about sensors and actuators. The true value of IoT lies in the digital connections it creates. Consider that before the internet was invented in the 1960s, digital information wasn’t very useful because it was all stored locally on individual computers and couldn’t be shared remotely. IoT connects information in a similar manner as the internet, but it connects data from operational technologies (OT) instead of information technologies (IT). By networking your assets, you can view the health of all assets in one singular, managed dashboard. Furthermore, a network creates many more data points than individual machines. Combining the OT network data, maintenance professionals can find similarities between machine failures and leverage a machine learning platform, which can improve the predictive algorithms over time.
  • SPdM automates some maintenance tasks. While predictive maintenance (PdM) can predict potential failures on machines, SPdM goes one step further by automating some maintenance tasks using cognitive computing technologies. For example, if a potential failure is detected on an asset, smart predictive maintenance might trigger a maintenance work order, assign a technician to it, and schedule to ticket in a computerized maintenance managed system (CMMS). Next, SPdM might check the spare part inventory required to replace the failing component in an enterprise resource planning (ERP) system and peg it to the work order. If the appropriate part is not on hand, SPdM could create a purchase request in the ERP system that only needs to be approved by the procurement specialist when they log into their account.
  • SPdM is integrated into other maintenance management systems. In order to automate certain tasks, the predictive maintenance platform should integrate with either the CMMS, ERP, or manufacturing execution system. Traditional PdM can fail if it is not built into your daily maintenance procedures. By integrating SPdM with other maintenance systems, you can gain both a sustainable platform and processes as well as the potential to automate more maintenance processes over time.

Five steps to reach smart predictive maintenance

Once the maintenance strategy and processes have been defined and implemented successfully, it may be time to start thinking about adding technology accelerators to your maintenance program. Although technology can, in fact, fast-track your maintenance program, world-class maintenance isn’t generally built in a day. The key is developing a thorough end-state vision and starting small with a pilot phase. Figure 2 lists the five steps you can take to reach SPdM.

Figure 2: Five steps to smart predictive maintenance

A pilot phase should generally take three to four months on a single production line or maybe just one or two well-suited assets and can cover the first two steps, asset monitoring and health and condition-based monitoring. The reason you generally can’t go straight to predictive maintenance is that it takes time to set up data collection processes, connect sensors to the machines, and most importantly, the asset has to fail at least a few times in order to baseline the predictive algorithms. The more a machine fails over time, the better the predictions should logically become.

Once data can be reliably collected remotely and an asset has provided enough failure data, the failure thresholds can be optimized. Then, a data scientist well-versed in statistical analysis can begin creating predictive models, measured by their degree of accuracy. Predictive maintenance can often improve machine availability and uptime by 10–20 percent.2 As more failures occur, a machine learning platform could monitor the failure data and update the algorithms to increase the predictive capabilities with each failure until unplanned downtime is all but eliminated. Keep in mind, this can take a while, which is why it’s essential to start early and focus on your most critical assets at first.

Finally, for those with a proper long-term vision, the last stop in this five-step process is SPdM, which we described in detail in the previous section. In the next section, we’ll discuss real-world use cases for some of the technology accelerators and how they apply to SPdM.

A day in the life of smart predictive maintenance

The following list details a fictional user story, as a maintenance supervisor uses SPdM to address an impending machine failure.

  • IoT: Machines are equipped with external sensors, which measure the temperature, vibration, and electrical current. The data is transmitted wirelessly to a remote private company server.
  • Cloud computing: The IoT machine data is aggregated on the remote server and an industrial IoT cloud platform visualizes the performance of each machine in the network, visible on a mobile dashboard to the maintenance supervisor and plant manager.
  • Artificial intelligence (AI): A potential failure is identified on one of the machines, and the SPdM platform uses AI to create a maintenance work order and purchase request for the required replacement part.
  • 3D printing: The replacement part must be partially powder-coated, so the maintenance supervisor 3D-prints a custom fixture to hold the part while it is being painted.
  • Robotics: The SPdM platform sends a signal to one of the welding robots to self-configure its home location to ensure its polar coordinates are accurately calibrated.
  • Augmented reality: Finally, the replacement part arrives and the maintenance technician goes to replace the part. This is her first time performing this job, so she uses industrial augmented reality glasses to see step-by-step instructions overlaid on the actual equipment while she fixes it.

Journey to SPdM begins with a small step

Smart predictive maintenance is here today. Companies in the energy and aerospace industries have been using PdM techniques for years, and will likely be moving toward SPdM in the near future. As described above, world-class maintenance can be the key to improving a plant’s throughput, efficiency, quality, and safety, while cutting maintenance overhead and spare parts inventory costs at the same time. And more importantly, getting to SPdM is a journey that starts with one very small step in the right direction. No matter where you are in your maintenance journey, we can help. Contact Deloitte for your maintenance strategy and operations needs.

Insert Custom CSS fragment. Do not delete! This box/component contains code needed on this page. This message will not be visible when page is activated.

Site-within-site Navigation. Do not delete! This box/component contains JavaScript that is needed on this page. This message will not be visible when page is activated.

Did you find this useful?