Baxter, Nancy (1991, ASQC) Thomas J. Lipton Company, Englewood Cliffs, NJ
Experimental design technology is useful for determining possible sources of variation and efficiently estimating the effects of variation in two types of processes: continuous process and batch process. This paper shows how experimental design techniques can help eliminate color variation in food products.
Applying experimental design technology to both continuous process and batch process included the following steps: parameter identification, design, analysis, and results. Two types of parameters were identified for the continuous process: those that could be controlled by the process (control parameters) and those that were inherent in the process (noise parameters). The second step for the continuous process employed a twelve-run Plackett-Burman design. The third step, analysis, plotted and examined the data. The fourth step, results, enhanced the interpretation of the results with graphical displays.
Five processing variables were identified for the batch process. The design step employed a half replicate of a 25 factorial design plus 2 center points (18 points total). In the third step, analysis of variance was used to estimate the effects of factors and rank the effects from lowest to highest. The last step, results, used data from the center points to determine the adequacy of the model and develop a probability plot.
The spirit of cooperation demonstrated by the Production, Research, Statistics, and Quality Control groups contributed to the success of the experimental design efforts
Food products,Plackett and Burman designs,Rework,Statistics