region of high vanillin yield when both the amounts of Cu-
(II) and NaOH are increased, whereas the experimental
conditions is given in Table 9. The results from these followup
experiments are reported in Table 9 and Figure 9.
The obtained model has shown an excellent predictive
capacity, as illustrated by the comparison of the estimated
response surface in Figure 8 with the experimental results
of Figure 9. The best yield of vanillin was obtained using
the conditions for experiment entry 4 of Table 9, in which
yexp ) 7.2% and the yield predicted by the model is exactly
the same, ypred ) 7.2%. This result represents a 75% relative
improvement compared to the standard Borregaard procedure
developed from the Monsanto process.1
(C) Oxidation of LS from Low-Temperature Digestion
Process (LS1200). A model similar to (A) has been developed
in the pilot plant reactors, and the obtained response surface
model is graphically represented in Figure 10. Optimisation
experiments, based on this response surface (Figure 10) were
performed. The interpretation of this model is very different
from the obtained model for LS750/LS850. A general description
of the model is: