According to a survey conducted by the Aberdeen Group, “best-in-class” companies are increasingly utilising the Internet of Things (IoT) and Big Data to implement Predictive Maintenance (PdM) models to address and improve their top operational challenges.
We’ve found that PdM can:
- Reduce unplanned downtime to 3.5% – Amount of unscheduled downtime against total availability
- Improve Overall Equipment Effectiveness (OEE) to 89% – Availability x Performance x Quality = OEE
- Reduce maintenance costs by 13% Year over Year – Total maintenance costs including time and personnel
- Increase return on assets (RoA) by 24% – Profit earned from equipment resources through improved uptime
Any industry that operates machinery – manufacturing, transportation, building automation – can benefit from PdM enabled by sensor-generated data. The value over any other maintenance model is that PdM empowers maintenance and operation decision makers to predict when an asset will need intervention well in advance of its failure impacting personnel, operations or production. PdM provides the highest possible visibility of the asset by collecting and analysing various types of data to provide the following benefits:
- Identifying key predictors and determining the likelihood of outcomes.
- Optimising decision-making by systematically applying measurable real-time and historical data.
- Planning, budgeting and scheduling maintenance repairs, replacements and spares inventory. PdM comparison example
The following example illustrates the amount of time that it takes to detect a potential failure interval for each of the four maintenance models commonly used today. PdM enables you to save time and money by detecting the failure based on data sources before damage to the machine occurs, says Kevin Terwilliger, director, IoT Innovation, Dell.
We recommend following these six steps to PdM success:
- Establish the business case for PdM – To make the case for a PdM implementation, the focus should be on the unique problems that affect optimising operational and production impacts while managing risk. It’s important to understand what metrics the organisation is focusing on and which need to be improved. Consider these questions to identify key goals of your PdM project and ensure success:
- How can data driven decisions be integrated within the constraints of your existing maintenance practices?
- How would the assets’ failure impact personnel, operations or production costs? What does downtime cost?
- What critical assets are likely to fail? When and why do we believe they will fail?
- Identify and prioritise data sources – The increase in asset connectivity and use of smart devices may have generated large amounts of available data. It is not needed or recommended to address this whole universe of possible data. Instead, begin to predict failures on a single asset by focusing on the usable, existing data sources related specifically to it. Figure 1: “PdM motor vibration analysis example” illustrates how capturing the real-time data from just one sensor resulted in avoiding costly damage, downtime and emergency response. Below is a list of the various types of data sources available and where they are typically found.
- Collect selected data – The selected data may reside in disparate locations from a device at the network edge to the server room to the enterprise cloud, including sensors, meters, enterprise asset management systems, and supervisory control and data acquisition (SCADA) systems. An ideal PdM solution should be flexible enough to enable you to collect from all of these data sources to learn and continually make better, more informed business decisions.
- Determine where to run your analytics – Establish an advanced analytics foundation based on your specific operation. For example, Edge (or local) and Cloud analytics can be balanced to reduce the burden of streaming perishable PdM data on your cloud deployment. A distributed approach enables you to detect and respond to local events at the edge as they happen, taking action immediately on streaming data, while simultaneously integrating additional data sources in the cloud.
- Combine and analyse data to gain precise insights – Start by analysing available data to define the parameters of normal operation for a machine. This enables the creation of rules through condition monitoring for analysing the real-time data coming directly from machine sensors. With edge computing devices like gateways, analytics can happen as close to the machine as possible with the native I/O to collect data from industrial equipment and the ability to operate in harsh environments. After analysing the real-time data, add historical and third party data such as reliability models and logs to uncover meaningful correlations, patterns and trends with the anomalies generated by the real-time data rules, to signal potential failures. The patterns can be used to further refine your rules and offer actionable insights in real time.
- Take action – Turn insights into action by integrating an aggregated risk assessment for all assets into your operation through a single dashboard. For example, when a potential problem is uncovered edge computing devices can trigger an event that allows you to send out automated alerts to concerned parties, such as location, estimated replacement parts and recommended corrective action to avoid a catastrophic event. Then, by capturing wear characteristics data from the replaced parts, you are able to continuously refine your PdM models and learn from performance insights. Finally, explore additional uses for your PdM data such as automating inspection reports and enhancing component supplier evaluation.
Read more in Dell’s PdM solution brief here.
The author of this blog is Kevin Terwilliger, director, IoT Innovation, Dell.
Comment on this article below or via Twitter: @IoTNow_ OR @jcIoTnow