Dana

In automotive driveline manufacturing, NVH (noise, vibration, and harshness) performance is a critical quality gate — failures translate directly to rework, scrap, and delayed shipments to OEM customers. Dana's Toledo driveline facility experienced erratic, "wave-like" fluctuations at its backlash validation stations, with first-time-through rates collapsing from 85–90% down to nearly 50% without warning. No single production line or model was consistently responsible. Local engineers validated test equipment, confirmed bearing preloads, and verified gear contact patterns — all stable. Despite exhausting every internal diagnostic avenue, the failure waves continued, forcing unplanned weekend shifts, elevated scrap rates, and chronic operational instability that local teams had no path to resolve.

Dana's Corporate Senior Manager of Manufacturing Data Science deployed Acerta's LinePulse platform to connect data across two geographically separate facilities: the Toledo final assembly plant and the Fort Wayne Manufacturing Center, which supplies Hypoid gears for Toledo's center-section assemblies. The first step was digitalizing years of legacy diagnostic files from Fort Wayne's Hypoid gear testers — records that had never been analyzed at scale — and ingesting them into LinePulse as structured, queryable process data. With a unified cross-plant data layer in place, the team used LinePulse's genealogy tracing to map failed NVH results at Toledo back through center-section assembly data and upstream to individual gear machining signals at Fort Wayne. Once the leading indicators of NVH spikes were identified, both plants deployed shared run charts, SPC-style process monitors, and real-time alerts that flagged tolerance drift within minutes rather than weeks.

Connecting upstream gear machining data to downstream NVH outcomes eliminated the failure waves that had destabilized Toledo operations. Key outcomes:

• Sub-assembly FTT rose from a median of 80% to a daily average of 98%, replacing hundred-part failure waves with only isolated single-digit misses
• Final axle assembly throughput increased by 8%
• Rework volume, scrap rates, and unplanned weekend shifts declined materially
• Both plants now operate from shared dashboards enabling genealogy tracing and cross-plant comparison in minutes rather than months

As Brian Longartner, Senior Manager of Manufacturing Data Science at Dana, noted: producing near 98–99% FTT "is pure impact to the Dana bottom line."

• Quality failures at final assembly often have upstream root causes that single-plant diagnostics cannot detect — multi-facility data visibility is a prerequisite, not an enhancement
• Legacy test equipment (gear testers, CMM outputs) frequently contains years of untapped diagnostic signal; digitalization unlocks it for ML analysis
• Cross-plant component genealogy tracing is the mechanism that makes intermittent, wave-like defects traceable to their actual origin
• SPC-style monitors and shared run charts at both facilities are essential to sustain gains after root causes are corrected — without live alerting, tolerance drift will recur