Anonymous Global Cosmetic Manufacturer

In the highly competitive global cosmetics industry — where a handful of multinationals control nearly half of worldwide output — production efficiency is a survival imperative, not a differentiator. When a leading cosmetics manufacturer brought a new line online at its Dallas plant (one of 29 total lines), it expected throughput of 100–120 containers per minute for mascaras and lip glosses. Instead, the line logged over 2,500 starts and stops per 12-hour shift. Without any manufacturing intelligence infrastructure for root-cause analysis, plant management had no systematic way to identify where or why stoppages were occurring — leaving significant capacity and cost on the table.

Prime Controls, a Rockwell Automation Solution Partner for Control, Process and Information, deployed FactoryTalk Metrics software to instrument the new line end-to-end. The line runs on Allen-Bradley ControlLogix and CompactLogix programmable automation controllers (PACs), connected via EtherNet/IP. Previously, operators had to physically leave workstations to manually log downtime against barcodes — and short stoppages often went unattributed entirely. Prime Controls engineered custom logic to automatically assign every downtime event to a specific root fault. After installing FactoryTalk Metrics on a local server and migrating existing data to the FactoryTalk database, engineers analyzed the line in live production mode, calculated OEE for each work-cell station, and reconfigured buffer alarm thresholds on the labeler to eliminate unnecessary state switching. All performance data became accessible via standard web browser.

After implementing FactoryTalk Metrics, the line's per-shift stop/start count dropped 90% — from 2,500 to approximately 250. The plant also recorded a 15–20% increase in uptime and is saving $100,000 annually in service-contract costs and productivity losses. Qualitative gains were equally significant: plant management can now evaluate true machine efficiency and product-loss calculations in real time, and maintenance teams have a timestamped log of downtime events tied directly to specific machine alarm codes. Two root causes were identified and resolved quickly post-deployment — the bottle feeder was being understocked (losing ~60 minutes of production per shift), and the labeler's buffer alarm was misconfigured, causing unnecessary state transitions.

• Visibility precedes improvement: 2,500 unexplained stoppages per shift were reduced only after automated attribution of every fault to a root cause — manual logging had masked the problem entirely.
• Integration gaps are a hidden OEE killer: Existing bar-code-based downtime tracking required operators to leave stations, leaving short stoppages unrecorded and unresolved.
• Simple fixes can follow complex instrumentation: Once data revealed the bottle feeder was starved 60 minutes per shift, the fix required no engineering — just proper stocking discipline.
• OEE scoring alone is insufficient: Identifying that the highest-OEE machine was running at capacity while downstream machines were starved or blocked required granular, per-station visibility that aggregate metrics don't provide.
• Payback is measurable at the line level: A single-line deployment delivered $100K in annual savings, making the ROI case straightforward for scaling to remaining lines.