This article is an excerpt from the reportThe Value of Foresight: Generating Value Through Integrated Predictive Maintenance. A copy of the full report can bedownloaded here.
In Part One of this series, we looked at a maintenance maturity model and the steps in the journey towards more advanced maintenance processes.Here in Part Two, we discuss what it means to take a holistic, integrated approach to PdM.
A Holistic Integrated Approach to Predictive Maintenance
Integrating Across Systems and Organizational Functions
Predictive maintenance works best within a holistic, integrated approach—i.e. bringing together and integrating data, systems, and human expertise from engineering, manufacturing, field service, logistics, supply chain, HR, finance … and of course maintenance. The foundation for this approach is building and maintaining a ‘digital thread’ for each asset—i.e. a full-lifecycle, digitally-connected approach to asset management, connecting all of the data and systems for each asset, from concept/design to manufacturing, service, and ultimately end-of-life/recycle.
In the ideal world, design engineers provide input to service analytics; CAD design is used in combination with sensor data to help simulate and predict the stresses and wear on machines; PLM systems ensure accurate BOMs are being used with the exact as-built and as-maintained configurations; supply chain planners receive (from predictive maintenance algorithms) better forward-looking indicators of spare parts demand and are thus better able to optimize spare parts inventory replenishment; logistics and field service personnel and systems are able to more efficiently allocate resources (equipment and personnel) and more cost-effectively plan tasks, based on more accurate forecasts; maintenance and service professionals receive highly accurate information about equipment configurations, status, and reliable remote diagnosis of issues, enabling them to bring the right tools and parts with them the first time, increasing first-time fix rates; parts provenance and remaining life is more accurately known; and design engineers receive highly granular data on the usage, failure rates, failure modes, and likely causes across the entire fleet of equipment deployed, enabling them to improve future designs.
Assessing Current Capabilities, Data, and Systems
Few companies have achieved this entire ideal state yet. It is a journey that can be addressed in ‘bite-sized’ pieces, while keeping the ultimate goal of end-to-end holistic capabilities front and center at each step of the way. An assessment of current capabilities, data, and systems can be used to determine where a company will get the most bang for the buck. This should include a ‘data inventory,’ looking at what relevant data exists across the company and assessing the quality and completeness of that data. Shop floor, field service, and maintenance professionals should be involved in the data inventory project. They are best suited to identify which data are valuable and what kinds of insights might be gleaned from that data.
The data inventory exercise will reveal shortcomings. Often people think they have a certain set of data, when upon examination they discover the data is missing, or difficult to retrieve, or inconsistent (different incompatible formats, spottily collected, etc.), or inaccurate, or has other issues. Some of these data issues require process changes. Others may be fixed by changes to systems or implementation of new systems.
Systems and Platforms Involved in Integrated PdM
A variety of organizational functions, platforms, systems, and data sources are involved in creating a holistic and integrated predictive maintenance capability, as shown below.
Figure 2 – Holistic Integrated Predictive Maintenance
Systems comprising an integrated PdM approach include:
- CAD—Computer Aided Design systems provide precise design data that can be useful in downstream activities such as product configuration; developing, documenting, and visualizing service procedures; and enabling augmented/mixed-reality-based maintenance and repair capabilities.1
- PLM—A Product Lifecycle Management system can help in organizing and cleaning up product-related data, such as part numbers. Clean accurate product and component data is a prerequisite for the machine learning of predictive maintenance to work well.
- MES—Manufacturing Execution Systems help gather real-time data from production equipment. That data is vital to doing predictive maintenance on those machines. MES systems can also help in the execution of production schedules to allow the repairs that PdM identifies.
- EAM—An Enterprise Asset Management system is core to more advanced maintenance capabilities. It provides full-lifecycle management of assets and the management and execution of maintenance tasks. EAM is critical for asset-intensive, highly regulated industries. The EAM system is both a source of asset-related data for PdM, as well as the means to schedule and execute the maintenance and repairs.
- IoT-Internet of Things capabilities can augment thinly instrumented assets with sensors providing real-time data to drive the ability of machine learning algorithms to better predict failures and maintenance needs. An IoT platform that can ingest, organize, filter, and analyze enormous streams of real-time data is a foundational element of an integrated PdM approach. The value of an IoT platform is realized by the applications built on top of it, such as PdM, production monitoring, fleet monitoring, and asset management.
- FSM—Field Service Management systems provide for scheduling, dispatching, and execution of service and repair, driven by predictive maintenance. Mobile apps provide more error-free and efficient execution of predictive maintenance tasks and may also provide more accurate data for prescriptive maintenance to learn by understanding precisely what actions were taken.
- Service Parts Management—These systems optimize the placement of the service parts required for PdM tasks to be executed. More accurate service predictions (resulting from PdM) allow service parts optimization engines to keep less inventory. Increased predictability allows better advanced planning of repairs. Thereby, fewer spare parts are needed, kept in centralized pools, rather than having to scatter a lot more inventory across the network to rapidly deal with emergency repairs.
- Supply Chain—Supply chain systems have the overall picture of demand and supply, driving production and maintenance schedules. The advanced warning provided by predictive maintenance allows supply chain professionals to help identify the best time to do maintenance on production equipment, taking into account upcoming demand, supply, and production schedules. Supply chain systems also play an important role in ensuring the purchase and production of spare parts needed for repairs.
- Logistics/Transportation—These systems manage the delivery of parts and supplies for maintenance activities. They can do better optimization of resources when a PdM capability provides them with more advanced notification of upcoming maintenance activities requiring transportation resources.
- ERP/Financials—ERP systems contain much of the master data consumed in the predictive maintenance process, such as part numbers and asset-related data. They provide the financial data and framework for costing and prioritization of resources and activities.
- HR—Human Resource (aka Human Capital Management or HCM) systems often provide tools to match the service technicians with the right skills for the specific needs that predictive maintenance uncovers. They also provide the tools to recruit and train technicians and maintenance professionals, who are often highly specialized.
- AI/Machine Learning—Artificial Intelligence (in particular Machine Learning) is a core component of predictive maintenance, needed to extract insights and predictions from massive amounts and varieties of data.
- Data Lake—A data lake is needed to clean up, consolidate, and make useable all of the diverse sources of data needed by the AI/ML engines.
Demand Management for Maintenance
Predictive maintenance provides ‘demand management for maintenance.’ It changes the maintenance planning paradigm. Ideally, predictive maintenance analytics estimate the probability of failure (and specific type of failure) at different time horizons. Based on the criticality of the various equipment at a location and the probability and consequences of failure, maintenance can be scheduled in batches to make the most efficient and cost-effective use of resources. Furthermore, all of the supporting functions (such as supply chain/logistics, spare parts, field service planning, etc.) are given a longer range, more accurate forecast of expected demand, so they can prepare better, more appropriately, at a lower cost. By integrating all of these functions and systems together, workflows and resource usage are better optimized, while simultaneously improving uptime and OEE. This increases overall productivity from a given set of resources, ultimately improving Return on Capital Employed (ROCE).
The Value of a Holistic Integrated Approach to PdM
A holistic integrated approach involves integrating many different systems into the PdM capabilities. This enables PdM to get the data it needs, but more importantly allows more integrated automation of the supply chain and service functions involved in executing maintenance and repair driven by PdM. To this end, organizations should seek integrated suites that bring together all of these components — advanced AI/ML, IoT, Data Lake, PLM, MES, EAM, FSM, Service Parts Management, SCM, TMS, ERP, and HCM. To the extent those components are pre-integrated, from a single provider, it becomes an enabler to rapidly implement more advanced PdM capabilities.
In Part Three of this series we look at how asset-owning organizations are using PdM. We specifically look at how PdM is driving value in extractive industries (e.g. mining systems, oil platforms), manufacturing plants, warehouses, transportation fleets, and facilities.
_______________________________________________
1 For example, helping a repair technician using an augmented/mixed reality headset or phone/tablet to highlight, right on the actual asset being repaired, which screws or part to remove or attach next, and even showing how to do it, as well as overlaying instrumentation readings and other key asset information onto the asset. — Return to article text above
To view other articles from this issue of the brief, click here.