| OCR Text |
Show Our extensive analysis of various statistical parameters of flame frequency spectra in different frequency bands, based on the "eddy concept", has brought us to a conclusion that the temporal flame frequency spectrum we measure and observe presents a distorted image of the real distribution of the fluctuational energy in the flame; and the degree of this distortion is frequency-related. In order to extract useful information from the chaos of the flame, we must take into account the effect of spatial averaging taking place in the flame. A proper analysis of the temporal frequency spectra should take this phenomenon into account and should assign different weight to parameters derived at different frequencies. On this basis, we developed a series of new algorithms related to burner stoichiometry and mixing rate which can be used for flame characterization and optimization. We also tested different approaches to forming optimum algorithms to characterize specific parameters, such as NOx and burner fuel-to-air ratio. We believe that our approach leads to a new method of combustion optimization with a broad range of commercial applications. This approach offers a simple way of practical implementation: raw flame signals from the existing burner flame scanners are connected to a computer which processes them and presents results to the operator and to the control system. Figures 2 and 3 illustrate the architecture and organization of the proposed system. Currently, we are working on testing our method on utility and industrial boilers equipped-with different types of burners, with the particular emphasis on 10w-NOx coal burners. Our new burner flame monitoring system will have the capability to automatically adjust to a specific burner design, type of fuel and operating conditions. Currently, the new system is packaged as a separate software which can be made compatible with existing Burner Management Systems. Our overall objective is to develop this new method into a system of combustion diagnostics and optimization which will allow to balance and optimize operation of individual burners. We are confident that the application of our new system will lower NOx emissions and will improve performance of commercial combustion systems. Key Experimental Results The first task of this project was concerned with collecting sufficient field data to evaluate the feasibility of the proposed approach. Our extensive data collection activities consisted of two parts: testing of utility boilers in a multiburner environment and at single-burner test facilities. Two types of flame sensors were used in our tests: a silicon flame sensor sensitive to mostly visible light and a lead-sulfide sensor sensitive mostly to IR radiation. Wherever possible, we utilized raw output signals from the existing burner flame scanners. In our tests, along with the flame signals, we recorded all available boiler, furnace and burner parameters, such as the boiler and burner loads, fuel and air flows, swirl vane positions, flue gas analysis ( O2, NOx, CO), furnace exit gas temperature, unburned carbon, etc. Testing at utility boilers had the main purpose to confirm that the temporal frequency spectra derived from burner flame radiation provide a reliable and reproducible measure of important flame characteristics such as stoichiometric ratio, NOx and flame stability. To date, tests were conducted at 3 |
