Equipment service engineers at one of our manufacturing customers spent around 80% of their time on solving escalations from unscheduled machine downtime. These disruptions are costly, unpredictable, and time-consuming – hampering scalability and growth.
To prevent unscheduled downtime, our customer introduced a special team: the “Proactive Escalation Monitoring (PEM) team”.
The goal: to move from unscheduled downs to scheduled downs (USD2SD).
To support the PEM team in their USD2SD quest, we built a highly secure Self-Service Analytics (SSA) platform, offering trusted datasets, and a plethora of tools for analysis, and visualization capability.
The platform provides a Mendix front end, and offers processing tools such as Databricks, Knime, Matlab and Python. Visualization is made easy with PowerBI, Grafana and Spotfire. The tools enable the team to quickly build and test models based on actual escalations and machine sensor trends. These models contain e.g. failure modes and lifetime of components and trigger field engineers to prepare for scheduled maintenance
The PEM team gathers field data by using an “escalation afterlife tool” in which service engineers document findings and solutions for each escalation, and in which they share ideas on how to proactively monitor machines. This information is fed into the SSA platform, to enrich datasets and create triggers that field engineers trust.
Field engineers use the triggers to prepare for swaps, act on sensor drifts, and foresee machine failures.
The SSA platform has helped our customer reduce the time needed for analysis, and it has prevented several unscheduled downs already.
PEM Engineer: “By using SSA instead of manual analysis, we’ve reduced diagnostic times from 72+ hours to minutes”.
The SSA platform currently has over 10 models running. From simple mathematical models to Machine Learning (ML) pipelines. Within six months these models have sent 53 triggers to field engineers, who managed to prevent 26 unscheduled downs with it!
With a growing number of PEM models implemented, the service engineers have more time for enabling proactive monitoring, preventing downtime, and saving millions of euros.