A Study on Beta Glucan Production Optimization through a 2k Saturated Factorial Experiment

In this paper, we consider the 2k saturated factorial experiment design to find the optimal point of production for beta glucan. A 2k saturated factorial design does not allow the replication of any design point, thus the error term has zero degree of freedom, making it impossible to test statistically all possible model term. In this case, if only main effects are included in the model, statistical tests can be performed, but the powers of these tests may be weak. Here, it is recommended to use the normal probability plot in order to find the statistically significant factors. In this study, the significant factors were found first using the normal probability of the 2k saturated experiment data. Then, in order to find the optimal point of production for beta glucan, an orthogonal polynomial regression model with the significant factors were used to estimate the optimal point within the experiment conditions. Furthermore, the optimal point was re-verified using the Taguchi method.