Track: Quality Control and Quality Management
Abstract
PT. S is manufactur company which produce electronic goods. One of the products which produced by company is a refrigerator with various types, which is 1 door, 2 doors, and freezer. The highest demand by consumer is the 1-door refrigerator. However, based on production data in month July 2017, from 69,612 1-door refrigerators produced, there were 5,577 defective refrigerator units, or as many as 8.012%, still far against zero-defect target. The research aims to analyze the problems occurring in the production process, and to propose quality improvement through the six sigma method with DMAIC approach and data mining to minimize the number of defective units. In the Define stage begins with the manufacture of Supplier Diagram, Input, Process, Output, Control (SIPOC). Followed by Measure stage which begins by doing observation to know Critical to Quality (CTQ), making Control Charts-U and Pareto Diagram. Based on the Pareto Diagram, the most dominant defects are the R Room Temperature NG defect, then calculate the Defect per Million Opportunity (DPMO) value and sigma levels. In the Analyze stage, Ishikawa Diagram and Five Why Diagram are created to find the root cause of the problem, then create FMEA's Table to determine the priority of improvement. Improve stage start with Design of Experiment 23 and replicate 5 times to find out the optimum setting. Known that optimum settings are temperature 710o C, 5 seconds weld time, and direct vise grip release. Followed by the stages of data mining for creating decision tree based on entropy and gain values that produce the ‘If Then Rule’ function. Based on the decision tree, Graphics User Interface (GUI) model can be made to know the decision of 'Good' or 'No Good' based on the numbers entered. Then, the quality improvement is given in the form of Standard Operating Procedures (SOP) which contains results from optimum settings and data mining. And at the last stage of Control, it is known that the percentage of R Room Temperature NG defects decreased from 11.99% to 3.303%, and the sigma level increase from 3.9992 to 4.056.