The Challenge
A corrugated packaging plant with four rotary screw compressors experienced unplanned breakdowns 8–10 times per year. Each triggered emergency maintenance, production stoppages averaging 6 hours, and expedited spare parts at premium rates. PM schedules existed on paper but were based on calendar intervals bearing no relationship to actual machine condition or load.
The cost of reactive maintenance extended beyond direct costs — production schedule disruption, overtime to recover lost output, and the accumulated stress on the team from operating in permanent crisis mode. Two technicians had left the company in the previous year; in exit interviews, both cited compressor breakdowns and the resulting pressure as contributing factors.
What Changed
Continuous monitoring of vibration, temperature, pressure differential, motor current, and oil pressure on all four compressors. Alert thresholds set at 80% of historical failure signatures. Automated work order generation on threshold breach.
Within three months, two compressors had triggered alerts — one for rising discharge temperature (aftercooler fouling), one for increasing vibration amplitude on a bearing assembly. Both were investigated and repaired during scheduled downtime windows. Neither progressed to failure. In the following 12 months, unplanned breakdowns dropped from 9 to 1.
Results
crisis-mode eliminated
“Predictive maintenance is not a technology project — it is a data project. The compressor already knows when it is about to fail: temperature creeps up, vibration increases, differential pressure climbs. What has been missing isn't the signal — it's the system to read it. When the signals become visible and actionable, reactive maintenance doesn't gradually improve; it nearly disappears.”