How is linear programming used for refinery planning and optimization?

Refinery LP models optimize crude selection, operations, and product slate. Structure: decision variables (crude purchases, unit rates, product sales), constraints (material balances, unit capacities, product specifications, crude contracts), and objective function (maximize margin = revenue - costs). Sub-models: crude blending, unit yields, product blending. Typical model has thousands of variables and constraints. Uses: monthly planning (crude selection, production targets), operations optimization (day-to-day adjustments), investment evaluation (what-if analysis). Shadow prices indicate value of relaxing constraints. Recursion handles nonlinear blending properties. Major tools: PIMS, RPMS, GRTMPS. LP is foundation of refinery planning; supplemented by nonlinear optimization for detailed operations and simulation for accuracy verification.