Response surface experiments on processes with high variation

Abstract

In industries such as food engineering, pharmaceuticals, bioprocessing and agrochemicals the use of biological materials means that, when experiments are being performed, run to run variation is typically higher than in many other engineering applications. This means that some of the smallest response surface designs are inappropriate and the traditional design principles of randomization, replication and blocking are more important. Recent work for response surface experiments of this type are reviewed in this chapter. Topics covered include the choice of run size, three-level designs, including some qualitative factors, arranging designs in various block structures, restricted randomization leading to multi-stratum designs and issues in the analysis of blocked and multistratum designs.