By Jonathan Gross
By Jonathan Gross
June 19, 2019 – Does unplanned equipment downtime cause your maintenance team to scramble?
How much do unexpected outages cost your company? What’s the value of delayed or cancelled customer orders?
According to a 2018 GE Digital survey, an average unplanned downtime event lasts four hours and costs US $2 million. Unplanned downtime also erodes customer trust. Forty-six per cent of respondents said that unexpected outages caused them to miss their customer order promise dates.
Exposure to unplanned downtime is pervasive. Seventy per cent of respondents said that they’re not fully aware as to when equipment is due for maintenance, upgrade and replacement. And, in our firm’s experience, maintenance departments don’t often reassess long-established static maintenance schedules. When services are scheduled too far apart, companies increase the risk of unplanned downtime. When services are scheduled too close together, companies incur avoidable parts and labour costs. It’s a no-win situation.
If unplanned downtime events were anomalous and difficult to treat, companies would have to live with the impacts. But many are not. Most reactive maintenance events are predictable. Parts wear over time. Output slows. Product quality begins to suffer. With routine monitoring and analysis, companies can do a better job of predicting failure, setting maintenance schedules, and planning part, people, and tooling resources.
With an industry-wide focus on delivering exceptional customer experiences, manufacturers can no longer expose themselves to unplanned constraints on production capacity. This is why, as a foray into Industry 4.0, many are setting their sights on integrated predictive maintenance.
Building your predictive maintenance technology architecture
For many, the big picture end-game is an environment where systems optimize machine performance, predict failures, suggest planned maintenance work orders, and pull the impacted machine resources from available master scheduling, material requirements planning and production capacity plans.
To pull this together, machines need to be connected to sensors and control systems. The data that’s generated needs to be consumed by artificial intelligence (AI) and enterprise asset management (EAM) systems. The EAM system needs to be able to create suggested maintenance plans based on the AI system’s predictive analysis and perhaps initiate changes to process controls. And, the ERP system needs to be able to incorporate EAM maintenance plans into its production scheduling and capacity-planning engine.
Planning your predictive maintenance strategy
Here are six tips to develop and execute your integrated EAM strategy:
1. Break down silos and build cross-functional teams. Traditionally, maintenance departments operate in silos of their own, separated from the rest of operations and the business. If your business is serious about integrating maintenance into the enterprise, it needs to take a holistic approach that covers organizational structures, business processes, data and systems. All impacted business functions should be involved. These often include maintenance, IT, finance, purchasing, inventory, and planning and scheduling.
2. Build an integrated technology architecture. Your integrated technology environment should consider production equipment, control systems, sensors and IIoT connectors, cloud, AI and BI, business applications, middleware, and infrastructure. If the long-term play is smart maintenance automation, you need to think big. What components are needed? What specific roles should each piece of technology play? What’s your middleware strategy to tie the pieces together?
3. Build your implementation plan. You’re not going to be able to implement everything at once. Break the long-term architecture down into manageable projects. First, build a strong digital twin foundation. This means ensuring that your digital world is a mirror of your physical world. Make sure that key systems are properly implemented, and that data are timely and accurate. You can’t risk having key assets pulled out of available production capacity if required parts aren’t available to complete the work order.
4. Then, dip your toes into predictive maintenance. It’s important that you trust the system. And trust takes time to build. Today, would you be comfortable sitting alone in the back seat of an autonomous car? Probably not. But, would you be more willing to trust automatic braking or lane assist technology? Probably. The same psychology applies to systems. It’s hard to give up manual controls. So, wade slowly. When you first connect your equipment systems to EAM, consider establishing user-defined parameters to trigger suggested predictive maintenance activities. For example, set upper and lower tolerances on conditions that can be captured by sensors, such as: vibrations, temperatures, flow rates, output and pressures.
5. Once you’ve built trust, take the plunge. Even though you probably won’t use the AI engine until some time in the future, you’ll want to set it up early so that it has a strong base of data for machine learning. Make sure your AI system is collecting and analyzing all relevant machine, process control, EAM and ERP data. Eventually, let your AI system predictively suggest planned maintenance activities based on relationships and trends that it’s self-discovering. In the long-term future, and to fully close the digital-physical loop, you might eventually have your EAM system integrate to your process control systems for real-time optimization. For example, where equipment starts overheating, your EAM system could signal your control systems to increase flow rates.
6. Ignore risk at your peril. If you’re using IIoT connectors, you’re exposing your production equipment to cybersecurity threats. An attack can halt production and even lead to a catastrophe. It’s important to think ahead. As you plan your implementation, you should consider a lockstep deployment of a broad and deep defensive cybersecurity strategy.
Manufacturers are hungry for Industry 4.0 initiatives that can drive measurable efficiency and customer experience benefits. Integrated predictive maintenance can help in both areas, which is why many manufacturers rank it near the top of their priority lists.
A sound approach to planning and delivery of predictive maintenance is critical and worth repeating. Work cross-functionally. Architect a solid technology environment. Build a strong digital twin foundation. Dip your toes into predictive maintenance. Take the plunge.
Jonathan Gross is the managing director at Pemeco Consulting. He helps his clients architect and implement technology environments that integrate ERP with the edge.
This article originally appeared in the June 2019 issue of Manufacturing AUTOMATION. Read the digital edition.