Lamar, Jim L.; Zirk, Wayne E. (1991, ASQC) Union Carbide Chemicals and Plastics Company, Inc. (UUC&P), South Charleston, WV
This paper uses case studies to demonstrate the value of nested experimental designs, a tool that analyzes variances in chemical processes by providing a breakdown of where the variability occurs. The examples include a simple three-level nested design, which is a useful technique for settling disputes between manufacturing units and the laboratory, and a four-level nested design, which is a useful technique for quantifying variation if a production run routinely consists of three consecutive batches. The four-level nested design may be either balanced or staggered. The author also discusses basic rules of variance.Union Carbide Chemicals and Plastics Company, Inc. (UUC&P) uses several methods of analyzing nested design, including: (1) SAS, which runs on a mainframe computer, (2) several copies of STATGRAPHICS, and (3) a mainframe software program written by an employee in the mid-1960's which was converted to Macintosh Basic in 1987 and was balanced for balanced or staggered analysis. The Macintosh and IBM versions estimate variance components of staggered designs up to six levels.Nested designs offer an easy, powerful and flexible technique for identifying potential sources of variation in the chemical industry, especially in the initial stages of improvement projects.
Chemical and process industries,Nested experiments,Statistics