Dynamic optimization techniques are applied for the optimization of crystal
lization processes. These obtain promising results, especially for difficul
t industrial applications with significant heat effects, concentrated slurr
ies, and state constraints. Here we introduce some concepts that focus not
only on the optimization strategy but also on the practical implementation.
As a case study, we consider a batch crystallization process, which has be
en studied in the field. The dynamic model includes not only moment equatio
ns but also thermodynamic equations to make the model closer to practical o
perating characteristics. Significant differences between this research and
previous work are that we incorporate the heat-transfer components and con
trol directly into the model. After demonstrating in plant that the dynamic
model is valid in both model formulation and parameter identification, we
optimize this model. The objective is to maximize the final crystal size in
order to obtain the highest purity of the desired product. Here, we use th
e package DynoPC, which includes recently developed dynamic optimization st
rategies. The dynamic model, consisting of differential and algebraic equat
ions, is discretized using collocation on finite elements. The resulting no
nlinear programming problem is solved with a reduced successive quadratic p
rogramming algorithm. The results are then compared with those obtained usi
ng a maximum principle for minimum operation time and with previous plant o
peration profiles. The optimal results show important improvements as the m
ean size of the crystals is 50% larger than the ones obtained under origina
l operating conditions.