Increasing the oxygen in air

In chemical, petrochemical and related industries, separation processes are usually responsible for a major part of the production costs. In the separation of gas mixtures, adsorptive processes like Pressure Swing Adsorption (PSA) are being used by small and medium-sized industries, mainly because they have been found more efficient and economical than traditional separation methods. As part of her research project for a 

We start our analysis using the 32 runs of the 2 factorial design in duplicate as a training set to fit a regression model. After that, we verify how well this model makes predictions for the rest of the runs, performed at the intermediate levels (the test set). 

Fitting a model containing linear, second- and third-order interaction terms to the training set data, we obtain the equation 

Central composite design for studying laccase production 

An efficient model should be capable of making accurate predictions for the response values within the investigated region. The prediction errors, as we already know, should not present any systematic behavior and their size should be comparable to the size of pure error. Fig. 6A.8 shows a comparison of the residuals from model fitting with the prediction errors for responses belonging to the test set. It is clear that something is wrong. 

---