ABSTRACT
Most of the real world optimization problems in engineering involve several inherent nonlinearities in their model and often require solution of either nonlinear programming (NLP) or mixed-integer nonlinear programming (MINLP) problems [15]. The presence of discrete variables, bilinearities, and
MUNAWAR A. SHAIK1 and RAVINDRA D. GUDI2
1Associate Professor, Department of Chemical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi, India, E-mail: munawar@iitd.ac.in, Tel: +91-11-26591038
2Professor, Department of Chemical Engineering, Indian Institute of Technology Bombay, Powai, Mumbai, India, E-mail: ravigudi@iitb.ac.in, Tel: +91-22-25767231
non-convexities further makes it challenging to determine global optimal solutions to these problems, which has been one of the important research topics in the literature [1, 7-11, 16, 19, 28]. Optimization techniques can be broadly classified as deterministic (or traditional) and stochastic (or non-traditional) approaches. On one hand, most of the deterministic optimization techniques such as branch-and-bound, cutting plane, and decomposition schemes either fail to obtain global optimal solutions or have difficulty in proving global optimality; while on the other hand, most of the stochastic optimization techniques have critical issues related to either slower convergence, longer computational times, and/or difficulty in handling of discrete variables.
