Page 97 - Design of Simple and Robust Process Plants

82 Chapter 4 Process Synthesis and Design Optimization

Fig. 4.2. Process split-up into sections and subsections for optimization.

example there are two methanol recycles which are valued differently. The recycle
from the last section back to the first section is valued at market value, while the
recycle from the separation section of the first section is valued at incremental cost
and capital for this specific stream to be spend in the separation section. In this last
case the methanol conversion is low, and the value is used to optimize the conver-
sion of methanol.
These intermediate process streams are updated each time that more accurate
information becomes available from the sections.
At the beginning of the synthesis study data are used from the feasibility study.
After this, the data are updated intermediately based on the selected flowsheet and
updated DFCs.

4.1.5
Optimization Methodology

The optimization methodology is based on two approaches:

1. Process optimizations are MINLP problems that are time-consuming efforts,
initially to build robust models. In addition, they often lead to nonconvex
optimization problems that do not converge to a global optimum. Moreover,
as process systems of chemical plants have a strong nonlinear behavior, bifur-
cations and also multiple steady states often occur. Another limitation to
MINLP problems is that the superstructure selection is not a trivial effort
where alternatives are easily overseen. Despite these disadvantages, the qual-
ity of the MINLP problem solver and the superstructure selection will
become more mature.
2. The evaluation of optimized alternatives: this is applicable when the number
of alternatives is limited.