Assessment of Lagrangean decomposition for short-term planning of integrated refinery-petrochemical operations

We present an integrated methodology for optimal short-term planning of refinery-petrochemical complexes (IRPCs) and demonstrate it on a full-scale industrial case study under four realistic scenarios. The large-scale mixed-integer quadratically constrained optimization models are amenable to spatial Lagrangean decomposition through dividing the IRPC into multiple subsections, which comprise crude management, refinery, fuel blending, petrochemical production. algorithm creates virtual markets trading blends intermediate streams within seeks tradeoff in such markets, with Lagrange multipliers acting as transfer prices. best results obtained decompositions two or three achieving optimality gaps below 4% all provides tighter primal dual bounds than global solvers BARON ANTIGONE, also improves computed using piecewise linear relaxation strategies.