Digital transformation drives new era of predictive maintenance
Published: 22 June, 2020
David Bean, solutions manager at Mitsubishi Electric, looks at how digitalisation – through the rise of the Industrial Internet of Things and technologies such as machine learning and artificial intelligence – is delivering new insights about plant assets, so boosting the potential for predictive maintenance and heralding the arrival of prescriptive maintenance.
Predictive maintenance is growing in popularity as an approach that enables companies to reduce downtime and control costs. Increasing asset availability and maximising plant utilisation are significant KPIs for the digital transformation of industry. New predictive maintenance techniques are accelerating this trend, defining a new paradigm that is being labelled as Maintenance 4.0.
The impetus across all industries to boost performance and improve competitiveness is becoming more acute, particularly for companies operating in international markets. While there are numerous KPIs that can be assessed as a measure of how well a company is performing, it is perhaps Overall Equipment Effectiveness (OEE) and Total Effective Equipment Performance (TEEP) that will be the most familiar.
Their popularity stems not only from the picture they provide of the performance of plants and individual machines and assets but also from the opportunities they afford for continuous improvement. OEE factors availability, production performance and product quality into a simple equation that delivers a score against which companies can benchmark themselves. TEEP extends this with a score based on OEE and utilisation to derive an understanding of the true capacity of a plant or asset over a given time frame.
Of course, these benchmark KPIs are only of value if companies act upon them to drive up their scores but it is notable the extent to which downtime (both planned and unplanned) impacts on these metrics. Unplanned downtime through, as an example breakdowns drives down asset availability whilst routine periodic or preventative maintenance reduces utilisation in an attempt to improve availability.
It is for this reason that predictive maintenance is growing in popularity as a strategy for minimising not only unplanned downtime due to unexpected catastrophic failure but also avoiding any excessive routine scheduled checks and servicing.
Changing from what might be expected in a traditional maintenance regime to using preventative maintenance can represent greatly reduced costs. Predictive maintenance reduces downtime because, as is implied by the name, you are able to predict a failure ahead of time and plan corrective action accordingly. It also reduces maintenance costs because maintenance is only performed when it is necessary. Estimates of cost savings range from 10% to 40% compared with routine periodic or preventative maintenance.
It drives a reduction in the time taken for repair and overhaul as well. This is because the issue is noticed before it becomes a significant problem and any associated remedial work can be simpler and easier to fix. Further, it is estimated that predictive maintenance which catches problems early – and thus enables corrective action to be taken before excessive degradation of machine performance – can also reduce product waste by up to 20% by reducing the occurrence of asset breakdowns during production.
Market trends
A recent Frost & Sullivan report identified that while preventative maintenance is commonplace, predictive is not and as many as 34% of facilities are spending over 30 hours per week on scheduled maintenance. That is eating into the plant’s capacity and introducing cost in terms of periodic maintenance. Add to this the fact that the same Frost & Sullivan report estimated the cost of unscheduled downtime across the European market at some €45bn and it is easy to see why predictive maintenance is coming to the fore.
Industry is changing though and adopting predictive digital technologies that are underpinning a step change in the potential of this maintenance strategy. The digital transformation process is disruptive but it is built on the analytical interpretation of data that is collected from plant floor assets. This data is becoming increasingly easy to capture, while the tools needed to aggregate, filter and interpret it are becoming more widely available.
Predictive maintenance utilises data collected from each machine and assesses it in relation to the machine’s normal pattern of operation. Any minute changes from, or inconsistencies with, the baseline data will lead to subsequent alerts with a prediction of future failure, so that optimised maintenance work can be planned. This way, any damage or fault remains isolated, so other parts remain unaffected, and total equipment failure is avoided.
The digital transformation of predictive maintenance, the so-called adoption of Maintenance 4.0 principles, builds on condition-based monitoring technologies but employs a far wider range of linked and networked sensors and devices as part of the Industrial Internet of Things. This feeds a far more comprehensive set of data points into new aggregation and analytics technologies, dramatically increasing the power and capability of predictive maintenance. In addition, advances in machine learning and artificial intelligence also have a role to play in predictive maintenance solutions.
Data is collected from an ever-greater number of systems and devices. What is now important is how we harness that data to make meaningful decisions about the health of an asset. We can do this by utilising a combination of Edge and Cloud based solutions together with a set of analytical tools to analyse and interpret that data. In fact, in using an Edge based solution, we can also enable real-time decisions to be made on the operation and maintenance of assets around the plant.
These are all advanced technologies, available now. Mitsubishi Electric for example has long offered its Smart Condition Monitoring (SCM) solution to alert impending problems on rotating machinery, facilitating a low cost predicitive maintenance solution. Beyond this, Mitsubishi Electric can provide Edge based solutions with embedded analytical tools, including the use of a digital twin. This can be used for predictive maintenance strategies amongst many other benefits in developing a greater understanding of the performance of a “real world” asset against its digital fully optimised model.
In terms of the impact of artificial intelligence and machine learning, Mitsubishi Electric is embedding its proprietary AI technology into some of its product range moving forward. As an example a future servo range will have embedded predictive maintenance features for motor bearing failure detection and conveyor tension loss and their MELFA robot software provides predictive maintenance information on each individual axis of the robot.
Enter the realm of prescriptive maintenance
Asset maintenance strategies are still evolving as digital transformational technologies develop further and we now see the establishment of a maintenance regime that does not simply point to a need for future maintenance. It can also explicitly diagnose the problem and feed back in real time remedial action to extend the lifetime of the asset without significant impact to performance. This concept of using prescriptive analytics has resulted in a new maintenance paradigm, namely prescriptive maintenance.
These new developing Edge technologies can not only be used to provide predictive maintenance information but could be extended to include links to logistics, inventory and MRO systems. By doing so, they provide a cohesive overview and execution of maintenance activity that has only been available in the past through the implementation of costly bespoke MES solutions.
Thus, digital transformation is unlocking the potential of predictive maintenance, turning data into meaningful information that can help companies increase plant availability and asset utilisation. And there is more to come as we move into the realm of prescriptive maintenance, all helping companies to optimise production in an increasingly competitive age.
