Page 190 - Sustainability in the Process Industry Integration and Optimization

Co m b i n e d P r o c e s s I n t e g r a t i o n a n d O p t i m i z a t i o n 167

their transshipment model for Heat Integration within a mixed-
integer linear programming (MILP) formulation for structural
process optimization. This work had been further extensively
developed (Duran and Grossmann, 1986; Floudas and Grossmann,
1987a; Floudas and Grossmann, 1987b). Fourth, stochastic optimization
has become popular in recent years, applying genetic optimization
(Shopova and Vaklieva-Bancheva, 2006) and especially simulated
annealing (Kirkpatrick, Gelatt, and Vecchi, 1983; Faber, Jockenhövel,
and Tsatsaronis, 2005; Hul et al., 2007; Tan, 2007).
Optimization methods can be classified according to the
characteristics of the objective function, the decision variables, and
the problem constraints (Guinand, 2001). A simplified classification
scheme for optimization methods is illustrated in Figure 8.1.

8.3 Optimal Process Synthesis

A process network uses a given set of operating units to create desired
products from specific raw materials. The objective of PNS is to
identify the most favorable (optimal) network for accomplishing the
given tasks. The P-graph methodology is a graph-theoretical approach
to solve PNS problems.
8.3.1 Reaction Network Synthesis
Every reaction is a material transformation, which corresponds to an
operating unit when mapped on a P-graph. Similarly, the maximal

Constraints and Objectives Decision Variables

LINEAR NONLINEAR CONTINUOUS INTEGER

Nonlinear
Linear Programming Programming Integer Programming
(LPR) (IP)
(NLP)

Integer Linear Mixed-Integer Linear Mixed-Integer Nonlinear
Programming (ILP) Programming (MILP) Programming (MINLP)

STOCHASTIC DETERMINISTIC STATIC DYNAMIC

Incorporation of Incorporation of
Probability Functions Time Domain

FIGURE 8.1 Classifi cation of optimization methods (after Guinand, 2001).