A Guide to Move Along the Journey Towards Prescriptive Maintenance
Our last blog article described the typical intent and benefits of the different types of maintenance strategy:
- Schedule Based;
- Condition Based;
- Predictive and
In this article we focus on the information, resource and process required to move along the journey towards prescriptive maintenance, identifying the key enablers to a successful outcome.
The Railway Industry, with its highly regulated and safety orientated culture, has well established Preventive Maintenance processes which are generally orientated around a schedule-based approach. In addition, some progress has been made on the journey towards more extensive Condition Based Maintenance capabilities, and the rail sector has the ambition to further exploit this capability by adopting (and/or evolving) predictive/prescriptive maintenance approaches, though its deployment is currently limited to a relatively small number of suppliers, operators and applications.
Reactive maintenance or preventative maintenance using a schedule-based approach doesn’t necessarily use ongoing knowledge of asset health and usage to assist in the maintenance decision process. Such processes only require limited knowledge about the current status of the asset – if the asset is available and operating successfully, and its service interval has not been exceeded then it is ‘business as usual’, and no maintenance activity is necessary.
In order to move along the journey towards prescriptive maintenance the first thing to recognise is that a successful prescriptive maintenance programme is dependent on a robust condition based maintenance at its foundation. Reliable condition monitoring is the source of asset health data that enables the advanced diagnostics and prognostics required for predictive/prescriptive maintenance approaches to be achieved.
Assuming the starting point is reactive or schedule based maintenance practices, what are the key enablers to facilitate climbing towards prescriptive maintenance?
Prescriptive Maintenance Key Enablers:
Reliable Condition Monitoring Data Prescriptive maintenance
Knowledge associated with the current and historic health (or condition) of the asset is essential. The use of condition monitoring data to assess the condition of an asset enables a proactive approach to maintenance to be undertaken. Condition Indicators are the essential element for this – parameters which are influenced by the health of the machine over time. Condition Indicators are derived from asset sensor data (typically temperature, pressures, vibration, speeds etc) to create an historical record of asset behaviour and characteristics.
The operating regime is defined by a set of conditions which, when satisfied, characterise a unique operating mode or running condition (the ‘regime’). Regime recognition is a function which monitors operational parameters to determine the operating regime of the machine, which can then be used to determine the appropriate application of condition indicators. Operating regimes are typically defined using a subset of operational parameters such as modes, speeds, torques, loads, power, direction, control positions, etc. For railway rolling stock applications, a selection of these parameters should be available from the TCMS.
Secondly, when monitoring fixed assets associated with track, it is essential to have awareness of the speed of the measuring train and geo-location of the track topology, including the various track elements or components (Insulation Joints, Switches, Bridges, etc). These can be identified as ‘Points of Interest’ that may have an influence on the parameters being monitored, enabling correlation with interesting signal characteristics such as sources of shock and other phenomena.
Situational Awareness for Prescriptive Maintenance
There are two main aspects to situational awareness for railway applications – regime and location.
Firstly, Condition Based Monitoring and Predictive Maintenance benefits are significantly enhanced if there is a mechanism to characterise the operating mode of the asset being monitored using regime recognition. To ensure high quality machine health data for analysis and thresholding it is beneficial to acquire and analyse the data during known (stable/defined and therefore repeatable) conditions, i.e. a fundamental requirement for reliable and meaningful trending data is the need to capture health data under known or consistent operating conditions, acknowledging the fact that the measured condition indicators (health parameters) of a monitored machine component will vary with the operating regime of the machine, as this affects loads on the component. For example, there is limited value comparing data collected at an ‘idle’ condition to that collected under ‘full operating power’ or ‘transient’ conditions, similarly ‘loaded’ versus ‘unloaded’.
Relevant Data for Prescriptive Maintenance
There is a common misconception that ‘Big Data’ is the enabler to predictive analytics. It is not. Vast amounts of data can be collected from the control and condition monitoring systems associated with an asset, but value can only be achieved if the data contains useful information. A more appropriate term is ‘Relevant Data’. There are several stages required to identify and isolate the relevant data. Firstly, data reasonableness/quality checks are essential to deal with artefacts such as noise, discontinuities, sensor issues, etc… this will help to minimise the risk of a ‘rubbish in = rubbish out’ scenario. Secondly, data pre-processing will likely be required to transform raw data into suitable streams for the creation of condition indicators. This may involve resampling, data aggregation, statistical analysis, combination of parameters, etc… Finally, the reliable condition indicators must be supplemented by the contextual information of situational awareness. The resulting relevant data may represent a small portion of the overall data collected, but that small portion of data is where the real value resides.
Put another way, the application of condition indicators applied in context, using situational awareness, takes out the variability of big data and creates relevant data of higher value.
Maintenance Records for Prescriptive Maintenance
Knowledge of maintenance records and failure mode characterisation is essential to be able to close the loop of monitoring, forecasting and action. The integration of maintenance knowledge with condition monitoring data enables the use of predictive analytics to assess current condition with respect to previous failure experience, resulting in potential for prognostic forecasting in terms of Remaining Useful Life estimates. Ultimately, linkage can be created between diagnostic/prognostic outputs and the maintenance instructions required to eliminate or reduce any risk of failure – this is where prescriptive maintenance benefits are realised. Maintenance knowledge fed back into the monitoring process can be used to mitigate the impact of changing machine characteristics following maintenance.
Motivation for Prescriptive Maintenance
Probably the most important, and often overlooked, enabler is absolute support for the maintenance initiatives being undertaken in your organisation. An effective maintenance strategy needs an effective change management process to make it happen. It needs to be embraced at all levels within the organisation and must be reflected in actual process and procedures. The transition from reactive and schedule-based maintenance to a predictive/prescriptive approach should be treated as a business transformation activity with appropriate stakeholder involvement and designated project sponsors. Roles, responsibilities, and maintenance practices will be impacted, so inclusive behaviour and effective communication is essential at all stages. A key part of this motivation is a user experience (from the CM solution) that actually helps people to do their jobs efficiently, rather than making it harder or less interesting.
In summary, it’s the combination of those key enablers that will create the opportunity to achieve an effective journey towards a successful Prescriptive Maintenance initiative, maximising diagnostic insight and prognostic foresight capabilities.
When you buy SmartVision™ you are not buying just a product but also have access to the extensive experience and knowledge of the EKE team. This experience and knowledge identifies relevant data and incorporates condition indicators to provide reliable condition indicator data. This can be enhanced by regime and location data to provide situational awareness. Links to maintenance records can be included to provide data for future predictive/prescriptive maintenance.