Empirical Study of Foundry Efficiency Improvement Based on Data-Driven Techniques

• Main Authors: Kuentai Chen, Chien-Chih Wang, Chi-Hung Kuo
• Format: Article
• Language: English
• Published: MDPI AG 2021-06-01
• Series: Processes
• Subjects: bottlenect detection; statistical data analysis; process variation; productivity; casting
• Online Access: https://www.mdpi.com/2227-9717/9/7/1083

In this paper, a data-driven approach was applied to improve a furnace zone of a foundry in Taiwan. Improvements are based on the historical production records, order-scheduling, and labor-scheduling data. To resolve the bottleneck provided by the company, historical data were analyzed, and the existence of large variance in the process was found. Statistical analysis was performed to identify the primal factors causing the variance, and suggestions were made and implemented to the production line. As a result, daily production increased steadily to more than 30 pots of molten metal, while the original production was 20–30 pots of molten metal and are not controllable. Such significant improvement was mainly made by standardizing the input and reducing the variance of processes. The average cycle time of each pot of molten metal was reduced from 219 min to 135 min. Our suggested improvements also reduced the foundry’s electricity consumption cost by almost $240,000NT per month. In summary, data analysis can help traditional industries in identifying the main factors causing the bottleneck.