Abstract
In order to improve the efficiency for solving MINLP problems, we present in this paper three computational strategies. These include multiple-generation cuts, hybrid methods and partial surrogate cuts for the Outer Approximation and Generalized Benders Decomposition. The properties and convergence of the strategies are analyzed. Based on the proposed strategies, five new MINLP algorithms are developed, and their implementation is discussed. Results of numerical experiments for benchmark MINLP problems are reported to demonstrate the efficiency of the proposed methods.
Highlights
Branch and Cut(B&C), Extended Cutting Plane Method(ECP)
MINLP models are usually classified into convex and nonconvex according to the convexity/nonconvexity of continuous model with relaxed discrete variables
From the solution of the NLP subproblem, assuming it is feasible, we obtain an upper bound of the primal problem, while solving the MILP master problem, a lower bound of the primal problem is obtained
Summary
Branch and Cut(B&C), Extended Cutting Plane Method(ECP). Reviews can be found in Grossmann, Belotti and Bonami[16,17,18]. There are several typical characteristics in process industries different from discrete manufacture industries, such as complex process networks, reaction mechanisms, fluid material flow that often involves nonlinear relationships among the flow variables. This has motivated the development of MINLP methods. We draw conclusions based on the performance of the proposed methods
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have