Central Composite Design Case Study

C. Central Composite Design
As PBD considers only main effects and ignores the interactions among the factors, therefore, a new design is required. Central composite design (CCD) is type of experimental design, which was first described by Box and Wilson (1951). Nowadays it is widely used in response surface methodology (RSM; discussed in the next section of the review) for building a second order (quadratic) model for the response variable without using a complete three-level factorial experiment. The design consists of three distinct sets of experimental runs (summarized in Table 5); first set is a factorial design in which the factors studied, each having two levels (+1 and -1); second set is a set of center points, where experimental runs
In order to maximize this response, path of steepest ascent is generally followed. Similarly to minimize the response, path of steepest descent is followed. Steepest ascent requires experiments according to a path of steepest ascent until the yield falls off. Path of steepest ascent is calculated after the initial screening experiment. Every point in the path represents an experimental run which is performed, continuing up the path until the yield falls off. To estimate the path of steepest ascent, it is needed first to fit a model. Typically a two-level factorial design (either full or partial) is used to estimate the optimum incremental changes required. Suppose an experiment has two factors and the interaction between factor x1 and x2 is not significant. Hence the model equation for this experiment
, 2003, Bajaj et al. , 2006). Another important limitation is the metabolic complexity of microorganisms. When a large number of variables are involved, the development of rigorous models for a given biological reaction system on physical and chemical basis is still a critical challenge. This is probably due to the non-linear nature of the biochemical network interactions and in some cases the incomplete knowledge about the kinetics involved in such systems (Franco-Lara, Link, 2006). Also, it is quite complicated to study the interactions of more than five variables and large variations in the factors can give misleading results possibly due to error, bias, or no reproducibility. To overcome the limitations of RSM another technique ANN has been started to be used by