Prescriptive maintenance is the execution of maintenance, that is “prescribed” or “recommended” by a system (i.e. Prescriptive System), that follows a repeatable method or process (Prescriptive Analytics).
Contrary to popular belief, prescriptive maintenance does not need a high organisational maturity to attain. Any system that makes a recommendation, is by definition a prescriptive system. If Maintenance is then performed on the basis of that recommendation, this is classified as "Prescriptive Maintenance".
Let’s start with the building blocks.
An asset modelling is foundational to a prescriptive maintenance system. An asset model contains all of the relevant information, knowledge and data required to understand an asset’s performance. The assumptions, data quality, and expertise contained within this model will have a significant effect on the quality of outputs from a prescriptive maintenance system. The asset model is represented by the following image:
An asset model can be fed information or data to inform a statistical prediction. This is the domain of predictive maintenance, and a typical use case would be to make use of sensor data as an input, to inform the current predictions. While use of sensor data is currently a "trending" field in asset management, It is worthwhile to note that you don’t need sensor data to be “predictive”. Up to date visual inspection information, ages, operational information, or any other available data can be used to alter the model predictions.
Understanding the current state of the asset, and then the future predictions (Performance, reliability, or otherwise), allows a prescriptive system to recommend appropriate actions. The determination of what is "appropriate", is typically its own defined method, such as cost/benefit, NPV of options or otherwise.
Analysis of the asset model, input data and causal relationships can be used to determine various paths forward and potential outcomes. These can be used to make decisions around the future operation, maintenance, and interventions at an asset level. Prescriptive analytics, and thus prescriptive maintenance, is an extension of the asset model and inputs configuration above. It now includes an element of analysis as shown below.
Since the analysis component is a method or process that can be coded and automated, it means that the analysis can be performed live or in a continuous manner, to produce recommendations and options. Any time an input to a an asset model updated, then the analysis can be re-run, and the recommendations and outputs updated.
Prescriptive maintenance is when the outputs and recommendation of prescriptive analysis are adopted and used regularly. A prescriptive maintenance approach ensures that an organization is acting on the latest "intelligence", and adapting their strategies to suit. Prescriptive maintenance is enabled by technology and software services such as modla.
In a prescriptive maintenance system, the prescriptive analytics is usually automated. The asset model becomes the area of focus for reliability engineers and subject matter experts, ensuring their current decision making and logic is captured correctly within the model structure.
A example configuration is shown below:
Modla's platform is built with the prescriptive analytics and prescriptive maintenance end goal in mind, however you do not need an abundance of data or high levels of organizational maturity to leverage the above approach.
Modla is an analytical engine, that when combined with asset models, provides live recommendations and analytics: