The operation

A pump system with standard SCADA-based monitoring. Multiple sensors covering pressure, flow, temperature, and vibration across the pump train and downstream process.

The challenge

The maintenance team was experiencing a pattern of unexpected failures on a specific pump component, despite operating within all configured alarm thresholds prior to each failure event. Post-failure investigation invariably showed that several alarms had fired in the preceding period, but they had been treated as independent events — logged, investigated separately, and closed without identifying a common cause.

The organisation suspected the alarms were related. There was no analytical capability to test that hypothesis on the historical data.

What Eyer found

Eyer's Fast Forward analysis on the historical data identified the failure sequence as a single correlated event. Four signals — pressure, temperature, vibration, and a downstream flow indicator — had been deviating from their learned baselines in a statistically correlated pattern in the hours before each failure event.

The primary deviation — the signal that deviated first, from which the correlated shifts in other signals followed — appeared five hours and twenty-five minutes before the operational impact. This was the primary insight: the failure was not sudden. It was a developing process, detectable in the data, hours before the threshold-crossing events that the existing system was configured to catch.

Alarm correlation in practice

The four alarms that the existing system had logged as four separate events were, in Eyer's analysis, one correlated sequence. The correlation was not coincidental: Eyer's time-series correlation model identified the statistical relationship between these signals from historical normal behaviour, and the deviation pattern was consistent with that relationship across multiple failure events in the historical record.

This matters for how operators respond. Four separate alarms require four separate investigations. One correlated alert with cascade context — what is primary, what is correlated, what the likely root cause is — requires one investigation, conducted earlier, before the failure has progressed beyond the preventable stage.

The business case

An early detection window of five hours and twenty-five minutes on a pump failure event is a substantial operational advantage. It is enough time to plan and execute a scheduled intervention during a convenient production window, rather than responding to an unplanned failure at the worst possible moment. The cost differential between planned and unplanned maintenance in industrial pump systems is typically measured in multiples, not margins.