Analyzing production efficiency in semiconductor manufacturing: A simulation approach using downtime data

Chai, Yu Qian (2026) Analyzing production efficiency in semiconductor manufacturing: A simulation approach using downtime data. Final Year Project, UTAR.

PDF Download (3367Kb)

Abstract

This study investigates production performance at an automated die testing line, known as the Multi-die Pick and Place Inspection (MPI) line, at Company X in Penang, Malaysia. Real downtime data from the MPI tracking system was incorporated into a Discrete Event Simulation (DES) model developed using WITNESS software, adopting a data-driven approach to overcome limitations of studies relying on assumed downtime values. The downtime data was analyzed and categorized into major error types to identify key factors affecting production efficiency. Overall Equipment Effectiveness (OEE), comprising Availability (A), Performance (P), and Quality (Q), were used as the primary performance metric. Results showed that Performance and Quality remained stable, while Availability was reduced due to frequent downtime, confirming it as the main limiting factor of system performance. A WITNESS-based simulation model representing 10 MPI machines was developed and validated against actual production data, achieving an average deviation of 0.02%, indicating high reliability. The model was then used to analyze operational scenarios, including throughput and revenue forecasting, product mix optimization, and performance improvement strategies. Results demonstrated a strong relationship between OEE, throughput, and revenue, highlighting availability improvement as the most effective strategy. Additionally, AI-based tools were developed to visualize results and support data-driven decisions. This study shows that integrating DES, OEE, and AI tools enhances production efficiency in semiconductor manufacturing.

Actions (login required)