| OCR Text |
Show Figure 9 illustrates the behavior of NO x, O2, CO and our flame quality parameters during deep changes in the position in the of SA and TA swirl vanes for the same single burner facility. The tests (series J) started with both SA and TA swirls at the maximum swirl (position 0), then the SA swirl was incrementally reduced to the minimum (position -220). NOx initially slightly goes up and then consistently goes down to the minimum. In series K, with the SA swirl vanes fully open, the T A swirl vanes were moved from the maximum swirl (position 0) to the minimum swirl (position -85) which caused a sharp increase ofNOx. Our results demonstrated that we can form certain flame quality parameters (qi) correlated with NOx, burner load, or even with the positions of T A and SA swirl vanes. We have also developed a different group of parameters (Si) which, in our opinion, provide a quantitative measure for flame stability. Our tests on single gas-fired burner have also brought us to some interesting observations. First of all, we observed the transformation of measured flame traces and our calculated parameters with frequency. This transformation allows us to form an output signal of any desirable, for control purposes, form: linear increasing, linear decreasing or extremum type. For the extremum-type curves, the extremum position in most cases corresponds to the position of "CO knee" in the "combustibles vs excess air" curve (Figures 10, 11). We know that in practical applications the CO knee corresponds to the optimum value of excess air. Therefore, this extremum can be used for burner optimization. The most important observation is that using the same approach, as applied to the coalfired burners, we are always able to form a very sensible, meaningful and repeatable response, although often hidden and masked in the chaos of the flame. This response may be correlated to NOx, fuel to air ratio and combustibles. Our tests on single-burner test facilities have reinforced our assumption that we can extract useful information from the chaos and randomness of the flame, and, what is particularly important, gave us hope that the path of this study may finally lead us to a truly universal method of combustion diagnostics, applicable to a broad range of fuels, burner designs and operating conditions. Future Development Our results indicate that temporal flame frequency spectra can be utilized as a practical basis for extracting useful information from the chaotic and noisy flame environment and specifically from the outputs of existing burner flame scanners and for developing a new ana effective method of combustion diagnostics. In the course of this project, we have tested several algorithms to form signals correlated with burner NOx, fuel to air ratio and flame stability.We were able also to make important steps towards practical application of our method and system. Based on results of our data analysis, we have formulated the following main goals and principles for the future development work: • universality (applicable to different fuels, types of burners and flame scanners) ~ • flexibility (compatibility with different types of measurements and controls) ~ • economy (no additional flame sensors or electronic devices around the burner)~ • reliability and ruggedness (immunity to many non-related factors). 11 |
