Effectively Applying Statistics to Accelerate Discovery and Improve Manufacturing Processes

• Activity Number: 214 - Contributed Poster Presentations: Quality and Productivity Section
• Type: Contributed
• Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Quality and Productivity Section
• Author(s): Wenyu Su* and Thomas Haynes and Jeffrey Wilbur
• Companies: DuPont de Nemours, Inc. and DuPont de Nemours, Inc. and DuPont de Nemours, Inc.
• Keywords: Experimental Design ; Statistical Modeling; Tolerance Analysis; Partial Least Squares; Generalized Regression; Regression Tree

Abstract

As a global leader in separation and purification technologies, the Water Solutions business at DuPont strives to manufacture best-in-class products. Various statistical techniques have been utilized by the business to accelerate discovery and improve manufacturing processes. Using three examples, this poster introduces how we have applied statistics in R&D research, pilot line studies, and manufacturing data analysis for process optimization and root cause investigation. Project objectives, key challenges, statistical deliverables and impact are discussed. Example 1 shows the power of experimental design and statistical modeling in R&D research for reducing experimental cost and speeding up the business decision process. Example 2 illustrates the value of tolerance analysis in pilot line study for reducing defect rate to guide manufacturing operation. Example 3 presents some challenges in manufacturing data analysis. Multiple statistical modeling methods including partial least squares, generalized regression and regression trees are used to mine production line data for reducing process variation.