| OCR Text |
Show propose to analyze primarily nonnalized signals and only factors characterizing the shape of the frequency spectra which should be the least affected by the non-related side effects. The important question pertaining to burner balancing is the following: if two burners are balanced in the abovementioned sense, does it give us the basis to claim that their air-fuel ratios are balanced as well? Although our flame sensor receives radiation and infonnation only from a small portion of the flame, we believe that continuous and intense processes of recirculation, swirling and mixing provide good averaging. Therefore, if we aim and install flame sensors on two burners identically, we expect to receive representative signals acceptable for balancing and adjusting these two burners. However, a more detailed and concise answer to this question can only be given after thorough testing of a single well instrumented burner when we could compare signals received from several different sensors aimed at different zones of the same flame. And finally, how does the wavelength sensitivity of a flame sensor affect the resulting frequency spectrum? Our test results have demonstrated that frequency flame traces from visible sensors have greater sensitivity and lower decay than traces from the IR sensors. This fact can be explained as follows. The IR sensors look "deeper" into the flame beyond the shear layer where the portion of smaller turbulent eddies and fluctuations are lower. The visible sensor receives its signal mostly from the surface of the shear layer where most of the eddies and mixing occur, and where the higher portion of smaller eddies, those causing higher frequencies, located. If this explanation is correct, then we should select and recommend visible sensors as the preferred type of the flame sensing device. However, our tests confinned, that for practical applications we had no difficulties in balancing burners on the basis of either visible or IR sensors. Our general observations and conclusions from the utility tests can be summarized as follows: 1. The utility tests confirmed our initial assumption that the temporal flame frequency spectrum, properly processed, exhibits a repeatable and reproducible reaction to changes in burner operation. 2. Comparison of flame frequency traces for the same burners in different series of tests has confirmed the signal repeatability. 3. The proposed manual burner balancing procedure, after a proper learning, and with proper equipment and methodology, can be conducted quickly, 5-10 min per burner. 4. Visible flame sensors have greater sensitivity to flame changes than IR sensors. 5. The concept of flame spectrum linearity proposed and patented by Fireye (Ref. 4) cannot be directly utilized for determining the "flame quality." Two flame traces, perfectly linear, can produce different unburned carbon and emissions. On the other hand, we observed very unlinear flame traces at good flame conditions. 6. Comparison offlame spectra for 10w-NOx burners with traditional high-NOx circular burners indicated a significant difference: the low NOx spectrum extends to higher frequencies and 8 |
