USE OF SUPERVISORY CONTROL TECHNIQUES FOR OPTIMIZATION OF ENVIRONMENTAL EMISSIONS IN POWER PLANTS

THIS PROJECT INVOLVES THE DEVELOPMENT OF A REAL-TIME SUPERVISORY OPTIMIZATION AND CONTROL SOFTWARE PACKAGE THAT WILL ASSIST PLANT OPERATORS IN RUNNING POWER PLANTS AT THE LOWEST POSSIBLE COST UNDER CHANGING CONDITIONS, WHILE MEETING ALL EMISSIONS REGULATIONS. COAL-FIRED POWER PLANTS ACCOUNT FOR 56% OF THE ELECTRICITY PRODUCED IN THE UNITED STATES, RESULTING IN EMISSION OF LARGE QUANTITIES OF PARTICULATES, SULFUR DIOXIDE, NITROGEN OXIDES (NOX), CARBON MONOXIDE, AND CARBON DIOXIDE, AS WELL AS SOLID WASTE PRODUCTS SUCH AS ASH AND FLUE GAS DESULFURIZATION SLUDGE. TO CONTROL ALL OF THESE SUBSTANCES WITHIN ACCEPTABLE LIMITS WHILE MEETING POWER PRODUCTION REQUIREMENTS AND MAINTAINING A HIGH EFFICIENCY IN THE BOILERS POSES A SEVERE PROBLEM IN CONSTRAINED OPTIMIZATION. THESE OBJECTIVES MUST ALSO BE MET IN A CLIMATE OF INCREASING COMPETITIVE PRESSURES ON UTILITIES. THE OBJECTIVES OF THIS PHASE I EFFORT ARE TO DETERMINE THE FEASIBILITY OF DEVELOPING AN ON-LINE SOFTWARE PACKAGE FOR MINIMIZING THE COST OF POWER PLANT OPERATIONS, TO SELECT THE BEST TECHNICAL APPROACH TO THIS PROBLEM, AND TO DEVELOP ACONCEPTUAL DESIGN OF THE SOFTWARE PACKAGE. IN PHASE I, SUPRVISORY CONTROL FORMULATIONS WILL BEDEVELOPED TO DEAL WITH A SELECTED SUBSET OF KEY POWER PLANT CONTROL INTERACTIONS INCLUDING THOSE BETWEEN NOX EMISIONS, COAL FINENESS, AIR DISTRIBUTION, AND THE COSTS OF ELECTRICITY. A NUMBER OF TECHNICAL APPROACHES INCLUDING ANALYTICAL AND STOCHASTIC MODELING AND NONLINEAR OPTIMIZATION, NEURAL NETWORKS, AND FUZZY CONTROL WILL BE EXAMINED BEFORE SELECTING ONE FOR USE IN SYSTEM DEVELOPMENT. A CONCEPTUAL DESIGN OF THE SOFTWARE WILL ALSO BE DEVELOPED IN PHASE I USING THE SELECTED CONTROL APPROACH, AND THE SYSTEM DEVELOPMENT WILL BE COMPLETED IN PHASE II. THE SYSTEM WILL BE DESIGNED TO BE EASY TO CUSTOMIZE THE DIFFERENT PLANTS AND EASY FOR OPERATORS TO LEARN AND USE.