Page 231 - Sustainability in the Process Industry Integration and Optimization

208 Cha p te r N i n e

capabilities: initialization of NLP subproblems; calling different NLP
and MILP solvers in a sequence with different option files (text files
containing specifications of solver options to be applied); the efficient
modeling of different formulations and strategies (e.g., multilevel
MINLP); the capacity to solve feasibility problems whose objective
functions are augmented by “penalties”; multiobjective optimization;
integer-infeasible path optimization; multiperiod optimization; and
flexible synthesis for cases where the true parameters are uncertain.
Some of these applications were described in Kravanja (2009).
MIPSYN can be comprehended and used at different levels of
problem abstraction because it includes: (1) an MINLP solver for
problems of a general nature; (2) a process synthesizer for generating
process flowsheets; and (3) a synthesizer shell for accommodating
applications from different engineering domains.
A number of case studies have been performed using MIPSYN.
In these studies, the synthesis was applied to all basic process
systems and subsystems. Examples include: (1) heat-integrated
reactor networks in overall process schemes; (2) heat-integrated and
flexible separator networks; (3) Heat Exchanger Networks, including
retrofits and networks that use more than one exchanger type;
(4) mass exchanger networks; (5) heat-integrated overall process
schemes based on a sustainable, multiobjective approach; and
(6) flexible and heat-integrated flowsheets, together with their
HENs, for cases involving as many as 30 uncertain parameters.
Note that the MIPSYN synthesizer shell also enables applications
in the area of mechanics (Kravanja, Kravanja, and Bedenik, 1998a;
Kravanja, Kravanja, and Bedenik, 1998b; Kravanja, Šilih, and
Kravanja, 2005). These mechanical applications range from simple
NLP optimizations to complex, multilevel MINLP syntheses of
structures in which topology, material use, and dimensions are
optimized simultaneously.

9.6.3 LINDO
LINDO is a tool for solving linear, integer, and quadratic program-
ming problems (Lindo Systems, 2009). It provides an interactive
modeling environment that facilitates the simulation and solution of
optimization problems. LINDO has the speed and capacity to solve
large-scale linear and integer models. The dynamic link library
(DLL) version of LINDO allows users to seamlessly integrate the
LINDO solver into Microsoft Windows applications that are written
in Visual Basic, C/C++, or any language that supports DLL calls.
Workstation users can exploit the linkable object libraries to hook
the solver engine to applications written in FORTRAN or C. The
latest LINDO version (ODC, 2009) offers a number of enhancements,
including: (1) significantly expanded nonlinear capabilities; (2) global
optimization tools; (3) improved performance on linear and integer