Focus Areas: Automated Process Analytics, Automotive Component Assembly, Root-Cause Engineering
The Executive Challenge: High-Volume Scrap and False Assumptions
In precision automotive component and electric motor manufacturing, press-fit operations represent one of the most failure-prone stations on the assembly line. When defect rates escalate, standard operating procedure often points to immediate, visible suspects: batch-to-batch material hardness discrepancies, tool wear, or operator loading variance.
On a high-volume assembly line, the operations team encountered a critical disruption: shaft pressing operations were producing an assembly rejection rate exceeding 343,000 PPM due to recurring radial shift variations.
Traditional line validation relied on scalar go/no-go limits—evaluating only the final seating force and total stroke displacement. Because parts frequently reached the designated final depth within the broad upper and lower mechanical limits, conventional inspection gates failed to catch out-of-spec components until downstream testing, resulting in compounding scrap costs and extensive rework overhead.
The Technical Bottleneck: Scalar Thresholds vs. Continuous Physics
Scalar thresholds evaluate the final state of an assembly, but they ignore the transition physics that occur during the stroke itself.
In a dynamic press operation, micro-binding, component tilt, or geometric non-conformance occurs within milliseconds during the initial engagement phase. If an assembly system only validates the terminal point:
- Early plastic deformation or component galling goes undetected.
- Misalignment during initial contact creates hidden radial stresses that lead to structural failure during dynamic motor testing.
- Standard MES dashboards report erratic, unclustered failures, leaving plant engineers chasing phantom machine faults.
Resolving systemic scrap required moving beyond static pass/fail flags to examine continuous time-series sensor signatures at scale.
The Analytical Intervention: High-Volume Trace Correlation
To identify the true root cause, the analysis bypassed aggregated line summaries and ingested raw cycle traces directly from the shop-floor Manufacturing Execution System (MES).
Raw MES Logs (100k+ Cycles) ➔ Automated Extraction & Normalization ➔ Force Curve Windowing ➔ Chamfer Redesign
- Pipeline Modernization & Trace Cleaning: Using custom extraction pipelines, over 100,000 discrete assembly cycles were extracted and converted into structured, query-ready datasets. Each cycle profile contained high-frequency data logging pressing force against stroke distance.
- Segmented Curve Windowing: Instead of looking at peak values, the press cycle was algorithmically segmented into three zones: initial contact (0–15%), linear displacement (15–85%), and final seating (85–100%).
- Anomaly Isolation: Clustering the time-series curves revealed a consistent, abnormal force spike during the first 12% of component travel. This initial resistance spike did not exceed the overall machine fault limit, but it directly correlated with radial misalignment downstream.
- Mechanical Pinpointing: Mapping the analytical stroke distance back to physical CAD tolerances revealed that a guiding chamfer had an inadequate lead-in angle. The shaft was catching on the edge of the bore before self-centering, forcing a skewed insertion.
The Operational Result: Permanent Defect Elimination
With conclusive empirical validation derived from production datasets, engineering teams executed a targeted mechanical redesign of the component guiding chamfer.
Simultaneously, the manufacturing line was configured with dynamic envelope limits—monitoring both displacement slope and force progression continuously rather than relying on terminal limits alone.
- Scrap Reduction: Assembly-line rejections dropped from 343,697 PPM to zero PPM, completely stabilizing the station.
- Cycle Assurance: Downstream motor cogging and dynamic performance test pass rates achieved stable statistical process capability.
- Systemic Visibility: The analytics framework established a repeatable template for diagnosing multi-variant assembly stations across the plant.
Consulting Takeaway for Plant Leadership
Advanced data analytics on the plant floor does not require multi-year enterprise platform migrations. High-ROI optimization happens when domain physics (tolerances, mechanical interfaces, kinematics) are coupled with modern data engineering to unlock insights already sitting dormant in your MES and machine logs.
Are your assembly lines validating process physics continuously, or are static pass/fail limits masking scrap on your floor?
About me
Shiwam Paraskar is an Industrial Data Architect and Manufacturing Analytics Consultant specializing in Industry 4.0 systems, MES data pipelines, and defect reduction for automotive and electric motor manufacturing. For diagnostic plant audits or targeted data automation initiatives, reach out directly at shiwamparaskar@gmail.com.