COAL COMBUSTION ACOUSTIC ANALYSIS UTILIZING NEURAL NETWORKS

LARGE MICRONIZED AND COAL-WATER SLURRY COAL-FIRED BOILERS UTILIZE MULTIPLE BURNERS FIRING INTO A COMMON FURNACE OR STACK, WITH THE COMBUSTION PROCESS BEING MONITORED THROUGH THE USE OF STACK SENSORS THAT ANALYZE THE COMBINED EMISSIONS OF ALL THE BURNERS. THIS COMPLICATES AND INCREASES THE COST OF ANY CORRECTIVE ACTION OR MAINTENANCE EFFORTS ON THE INDIVIDUAL BURNERS. BECAUSE OF THIS, THERE IS AN UNFULFILLED REQUIREMENT AND A MARKET NEED IN EMERGING CONTROL/MAINTENANCE TECHNOLOGY TO FINETUNE INDIVIDUAL BURNERS FOR OPTIMUM STOICHIOMETRY AT MINIMUM NITROGEN OXIDES (NOX). THERE IS ALSO A NEED TO QUICKLY CORRECT ANY OUT-OF-SPECIFICATION CONDITIONS. CURRENT TECHNOLOGY CANNOT ASCERTAIN FOR EACH OF MANY COAL BURNERS THAT FIRE INTO A LARGE BOILER WHETHER AN INDIVIDUAL BURNER IS FIRING AT OPTIMAL STOICHIOMETRY AND AT MINIMUM NOX. AN ACOUSTIC SENSOR BEING DEVELOPED IN THIS PROGRAM SHOULD BE ABLE TO FILL THAT NEED. PREVIOUS RESEARCH AND DEVELOPMENT (R&D) WITH ACOUSTIC SENSING SHOWS THAT THE SOUND POWER LEVEL BETWEEN DIFFERENT COMBUSTION OXYGEN LEVELS VARIES. ADDITIONAL R&D IS REQUIRED BEFORE THIS TECHNOLOGY CAN BE DEVELOPED INTO THE USEFUL PRODUCT FOR WHICH THERE IS SIGNIFICANT COMMERCIAL DEMAND. IN PHASE I, TESTS WILL BE MADE AT COAL-BURNING TEST FACILITIES (E.G., PENN STATE, ABB COMBUSTION ENGINEERING SERVICES). ACOUSTIC DATA WILL BE GATHERED OVER A WIDE RANGE OF BURNER CONDITIONS, USING ACOUSTIC WAVEGUIDES MADE OF MAGNESIUM OXIDE AND OTHER MATERIALS. THE WAVEGUIDES WILL BE USED IN BOTH THE CONTACT AND THE NON-CONTACT MODE. THESE DATA WILL BE ANALYZED, SEEKING CORRELATIONS WITH NOX LEVEL, EXCESS AIR, ETC., LOOKING FOR SIGNIFICANT PATTERNS.