OCR Text |
Show control valves in the auto mode. The burner air flow could be precisely adjusted with a manual damper. Three types of flame sensors were used in our tests: ultraviolet, visible, and infrared sensors. ~n all cases, we received similar results using different types of sensors; the results differed In sensitivity, but qualitatively were the same in all our tests. We conducted the tests at three loads: low load (LL) - 0.55 MBtuIhr, intennediate load (IL) - 0.72 MBtuIhr, and high load (HL) - 0.825 MBtuIhr. During the tests, we made step changes of the air flow at constant fuel load. We varied the air flow at the maximum possible range, far exceeding normal operating conditions. Of crucial importance when working with flame fluctuations are the measuring techniques and means of statistical analysis. To date, we have developed and tested several software packages specifically designed to process fluctuations of analog inputs. The latest packages include our new BDS flame-monitoring system (patent pending) and two related derivative systems: FlameTest (FT) and FlamePlay (FP). The BDS is designed for continuous burner-monitoring operation, including multibumer installations. It receives analog inputs from a data acquisition card, performs the necessary statistical calculations, and presents results to the operator in bar graph andlor strip-chart form. An important feature of the BDS is the formation of a specific algorithm which automatically accounts for changes in burner load and combustion conditions. The Ff system is designed specifically for testing. It presents all analog inputs and their calculated statistical values in the strip chart and X-Y plot forms and allows to record input data into computer. The FP system facilitates post-test analysis. It allows the operator to replay the records and to make any desired changes in the system calculations and adjustments. This software suite - BDSIFfIFP - is an effective tool for analyzing burner flame fluctuations to determine correlations between combustion parameters. We are convinced that fluctuations of most combustion parameters carry important and effective information. By applying proper measuring and signal-processing techniques, we can extract this ,information from the chaos of the industrial flames. Test Results Our first step was to formulate the technical requirements which the new system should meet. On consideration of an ideal burner diagnostic system, we concluded that such a system should generate an array of at least four types of output signals for each burner. Figure 1 illustrates the four types of signals. The Type 1 signal correlates with changes in air-fuel ratio or air flow to the burner. (Signals of this type usually provide good correlation with NOx.) The Type 2 signal correlates with changes in combustibles andlor unburned carbon. (This signal provides good indication of the CO knee.) The Type 3 signal correlates with burner load. 3 |