Lab-Scale Fiber Spinning Experimental Design Cost Comparison

Abstract

Many statistical experimental designs are too costlyor require too much raw material to be feasible forlab-scale fiber spinning experiments. In this study afour-factor response surface design is presented tostudy the fiber spinning process in detail at the labscale. The time, cost, and amount of raw materialrequired to execute the proposed design are comparedto the typical completely randomized 24 factorialdesign used in fiber spinning experiments and also toa standard four-factor response surface design.Sample fiber data as well as analysis from a typicalstatistical software package is provided to furtherdemonstrate the differences between each design. Bydesignating some treatment factors in the design ashard-to-change, split-plotting is used to reduce thetime, cost, and amount of raw material required tocomplete the experiment. The proposed split-plotdesign is faster and less expensive than a typicalfactorial design and has the advantage of fitting amore complex second-order model to the system.When compared to a standard response surfacedesign, the proposed split-plot design provides thesame second-order modeling capabilities but reducesthe cost of the experiment by 53%, the total time by36%, and the amount of polymer required by 24%.Thus, a split-plot response surface design based onhard-to-change factors is recommended in lab-scalespinning.