Screening and Optimisation of the Current Oxidation
Procedure
(A) Oxidation of LS from High-Temperature Digestion
Process (LS750 and LS850). Statistical experimental design23-25
and modelling using multiple linear regression26 and the
partial least-squares regression24 was used to obtain an
overview of the influence of the experimental, controllable
variables in the current procedure for the basic hydrolysisoxidation
of LS.
Since the current vanillin process has been run for nearly
30 years with concomitant process evolution and optimisation
studies, an overwhelming amount of process and experimental
data were available. These data were, however, not
consistent for modelling purpose, mostly due to changes in
LS qualities and changes in the hydrolysis-oxidation unit
of the plant. However, in the screening phase of the present
study, some of these process and experimental data material
were used as a support for selecting variables and experimental
levels for a screening study. Hence, the experimental
variables the reaction temperature (x1), the reaction time (x2),
the amount of air (x3), the amount of CuSO4 (x4), the amount
of NaOH (x5), and the amount of lignosulfonate (x6) were
all expected to affect the vanillin yields and were thus
investigated in a 26-2 fractional factorial design.
The experimental levels for these variables (Table 4) were
based on the existing standard oxidation procedure. Other
variables, such as reaction pressure and stirrer speed, were
also considered to influence the reaction, but it was decided
to perform the experiments at fixed levels for these variables
based on observations from earlier studies using LS750. An
empirical model, eq 10, was derived from the design matrix
and the analytical results (vanillin yield) given in Table 4.