LS Cable SCR Process Crack-Cause Analysis & Prediction Model

01 / Project note

I joined an LS Cable project diagnosing data readiness for SCR (continuous casting & rolling) wire-breakage factors and process conditions, owning data processing and analysis (30% contribution). I cleaned and analyzed process data to identify the causes and patterns behind cracks, and proposed manufacturing improvements with a baseline prediction model.

Wire-breakage cracks in the SCR process were hard to attribute to a single cause because multiple process variables were entangled, and the raw process data contained segments where casting speed deviated from normal operation — analyzing it as-is risked distorting the root-cause findings. I first built a filtering algorithm to remove casting-speed anomaly segments, securing the reliability of the analysis dataset, then cross-analyzed process variables against crack occurrence history on the cleaned data to identify patterns and control conditions strongly correlated with cracks. Based on these patterns, I built a baseline model estimating crack probability, proposing process-improvement directions for the manufacturing team to review.

• Improved analysis data reliability with a casting-speed anomaly filtering algorithm
• Proposed manufacturing process improvements with a crack-probability baseline model