Today, industrial companies' machinery is equipped with many different sensors that continuously generate data. Among other things, they can be used for predictive maintenance - the predictive maintenance of machines. If certain threshold values are exceeded, for example for component wear, employees receive an automatic notification about necessary maintenance windows or impending breakdowns at an early stage. Based on this, those responsible plan the replacement of consumables or components before a defect occurs. This allows MRO (maintenance, repair and operations) processes to be designed more efficiently, increases the availability and service life of machines and, above all, prevents downtime.

A study conducted by the International Data Corporation (IDC) on behalf of Intersystems shows that the majority (51%) of companies surveyed in Germany have already invested in predictive maintenance and intend to continue doing so in the future. Predictive maintenance enables companies to quickly find suitable answers to many questions. Solutions that enable prescriptive analyses promise to make things even easier. This is an approach that is based on AI and takes the concept of predictive maintenance a step further.

Prescriptive analyses in production

Prescriptive analyses not only provide valuable insights, but also give specific recommendations for the management of processes - with full transparency about the factors and effects of each individual alternative. This includes, for example, information on the time required and the costs of a measure. This approach leads to greater efficiency in planning and decision-making. Such solutions also offer major advantages in terms of machine management. For calculations and simulations, they access historical and current data from internal and - depending on the implementation - external sources. Recommendations for action are either given proactively in the form of an automated early warning system or called up by employees on request. They then weigh up the best course of action themselves based on the suggestions and subsequently initiate the necessary processes. Depending on the maturity of the solution and the AI model, sub-processes can also be fully automated if required.

Proactive recommendations for action

Prescriptive analyses are a supplement to the use of predictive maintenance. The AI-based recommendations for action enable employees to better coordinate maintenance management and associated processes, for example in the areas of resource planning and purchasing. If required, a solution for prescriptive analyses can suggest an optimal maintenance window and assign the right specialists to the upcoming tasks. All of this also takes place with predictive maintenance. However, prescriptive analyses have another major advantage: the solutions behind them order spare parts from their own warehouse or from other locations or initiate the necessary procurement process. Procuring these just in time optimizes warehousing. This means that only the spare parts that will actually be used in the near future are added to the stock. Overall, prescriptive analyses increase machine availability and utilization. Interruptions to production are minimized, costs are reduced and employees notice a noticeable reduction in workload.

Prerequisite: uniform and up-to-date database

Prescriptive analyses require direct access to all relevant production and logistics data. Companies are therefore initially faced with the task of linking data of any format from various sources. The second crucial step is to harmonize and normalize the very different data from OT (Operational Technology) and IT (Information Technology) systems. Data comes together from all departments that play a role in MRO processes. In addition to production, this also includes purchasing and logistics, for example.

A modern data platform is the key. It overcomes the boundaries of the many existing data silos. To ensure interoperability between internal and external systems, a modern data platform masters all common standards, protocols and profiles for data exchange in the industry. This creates a uniform and always up-to-date database for prescriptive analyses. Ideally, the data platform also has integrated functions for AI-based evaluations in order to achieve a fast time-to-value when introducing prescriptive analyses.

Whether on the store floor or in the back office - there are many other use cases for prescriptive analyses. Once asset management has been optimized, the experience gained can be seamlessly transferred to other projects. For example, a prescriptive analytics solution can also be used to optimize order sequencing. In addition, employees can anticipate and proactively address disruptions in supply chains. Based on the recommendations for action, it is possible to reroute shipments, request goods from warehouses at other locations or transfer orders to them, or find alternative sources of supply. Prescriptive analyses also help with demand forecasting.

Werner Reuß, Manufacturing Solutions Executive at InterSystems / red