In 1965, the industry introduced Computerized Maintenance Management Systems (CMMS), a software that streamlines maintenance operations by centralising maintenance data. Then, as maintenance operations matured, they incorporated adjacent business functions such as procurement and inventory management. The vertical integration of maintenance operations gave rise to Enterprise Asset Management (EAM) Systems which in the 1990s, further expanded into Asset Performance Management (APM) Systems.
Let’s compare the evolution of IBM Maximo’s with this history. The original Maximo transformed from a CMMS in the 1990s to an industry-aligned EAM system post-2000. IBM Maximo Application Suite extends the EAM paradigm and enables the move toward Asset Performance Management (APM) and Asset Investment Planning (AIP).
In the 1990s and 2000s monitoring and control systems used statistical process control methods to identify anomalous operating behaviours. These approaches required human evaluation to validate true alerts and disregard false alarms. Many organisations have asset performance management basics in place, many can schedule and classify asset upkeep, and most have monitoring and inspection regimes, but these qualities no longer define best-in-class asset management practices – they are the bare minimum. For an organisation to remain relevant, further abilities must be developed.
What is asset performance management (today)?
APM extends reliability methodologies of physical asset management by enhancing them with new technological enablers such as Artificial Intelligence (AI) and the Industrial Internet of Things (IIoT). In this way, APM is a new technology-based solution and an improvement to tried and tested practices.
APM solutions have a broad set of functional requirements as the market is constantly evolving. Traditionally, OEMs, monitoring and control systems and maintenance systems had distinct functions. With the advent of APM, each party had to expand their offerings. Equipment manufacturers are fitting monitoring devices and management systems into their products, Supervisory Control and Data Acquisition (SCADA) systems are integrating into maintenance systems, and EAM solutions are increasing data acquisition for advanced machine learning tasks.
Since each solution approaches APM from a different angle, the market is fragmented by geography, industry, and specialised asset classes. These approaches should converge into a reliability-centric solution that allows organisations to understand and act on the health of their assets.
Asset health and reliability
It has been proven that reliability methods and practices such as Reliability Centered Maintenance (RCM), Root Cause Failure Analysis (RCFA), and Failure Mode and Effect Analysis (FMEA) improve the overall reliability of assets. Historically, these methods have been implemented and managed manually, and data was acquired largely by visual inspection. A limited set of critical measurements is frugally observed using remote monitoring.
The onset of computer vision and cost-effective monitoring solutions reduce the barrier to entry, resulting in increased data velocity and dimensionality. A specific example is DES’s solution, which accurately extracts the relative movement of a captured object to a millimetre using a smartphone’s (or any other) camera. By using such technology, bridge bearings can be monitored for cyclic movement caused by temperature fluctuations. The increased movement would have resulted in premature failure of the bearing, however, as this was dedicated earlier on, the bearings life was extended.
APM formalises a scalable system supporting these processes by drawing Operation and Transactional (OT) data from an EAM system and infusing it with IT data generated from IoT devices and SCADA systems to provide a real-time view of asset health. As an example, an EAM system contains a comprehensive history of past work, future planned work, lifecycle information and movements (locations) of an asset. Such context can clarify why a failure occurred or when coupled with predictive analytics, could predict ahead of time what the failure will be.
What does AI do in an APM system?
From predictive forecasting to anomaly detection, AI and Advanced Analytics in the APM context act as the automated interpretation layer between the IoT device and the APM system. Some of the most prevalent uses cases in the APM context include:
- Anomaly detection: Sensors can occasionally provide erroneous readings, resulting in false positives and excessive investigations by the maintenance team. Anomaly detection models learn the behaviour of a given asset and filter out false positives ensuring data integrity.
- Feature extraction: Incoming data may correlate with a given event in the frequency domain, time domain or both. The frequency domain is concerned with the number of occurrences within a given range, and the time domain focuses on how the signal changes with time. Various algorithms exist to process sensing data in both domains, extracting a meta dataset and yield stronger correlations between raw inputs and actual events.
- Machine learning: Sensing data represents the behaviour of an asset within a given context. Machine Learning models are deployed for each asset to understand how it responds to a given input within its unique context, effectively mapping input sets to outcomes. This is applied to learn when a failure mode is imminent and can be applied to suggest corrective actions.
- Predictive forecasting: Leveraging anomaly detection, feature extraction and machine learning an asset operating condition can be predicted. If the next maintenance activity occurs after the predicted failure date, warnings can be issued with a suggested correction activity to ensure the asset does not fail.
The adage of working smarter not harder is the central purpose of AI. Identify the key parameters that define an asset’s health once, let a system continuous monitor them and provide feedback to users when they are not operating as expected. In this way, APM systems separate the vital few from the trivial many and divert attention, effort and investment at critical assets to maximise their return on investment.
The Implementation processes
Why is asset performance management needed?
Organisations with a growing asset base are trying to do more with less, more uptime, less work, more expert knowledge, less cost, more impact, less effort. Once the organisation reaches a Pareto optimal (a point at which many options are optimal), the problem becomes one of prioritisation; where must our effort be applied to maximise our contribution. Therefore, searching for an optimal point to apply effort requires understanding behaviour and how the equipment will respond under certain current and future conditions.
Human intuition used is the industry standard for inferring the behaviour of an asset, and for a good reason – it works for many observable phenomena; however, information experienced outside the realm of human observability is far more susceptible to error, such as the bridge’s movement under different temperatures or crystal deposition within a tube. The problem is worsened with scale; an increased asset base reduces the effectiveness of human observability either by introducing bias or diminished capacity.
As our Innovations Director, Tony Turner put it, Maximo enables the organisation to integrate operational, asset condition and locational data with cyber interpretation and analyses, ensures not only safety of customers and assets, but also the optimisation of value for both the company and its clients.
Download the latest IBM MAS Overview to learn more about all features and capabilities.
More blogs that you might be interested in Maximo Application Suite is no Rubik Cube: flexibility in a container
Share this Blog Post: