OCR Text |
Show 10 r-----,-----~------r_----~--__ o . - ..... -,.- ._ .. -----., ---- - • U o u' AW ~ w' . . ... ~ . ·l·A···· ~~iA·~-~·l~A~A·~·~···· • • . : : : A . ~ .. ~ ~- ........: - ........ :. . ....... ; ..... . A •••• ••• : : : • • -2 . -- .. -. : .. - .... . . : .••••••••• ~ .•• . . -4 '-----"----.l...-. __ ..I..--__ .L...-_....J o 5 10 15 20 Distance between probe tip and measurement volume [em] Figure 7: Effect of probe tip purge flow 011 measurement accuracy inlet swirl number = 0.6, purge flow outlet velocity = 5 m/s In conclusion, the lack of a perfect measurements repeatability is the main source of inaccuracy for LOV measurements in swirling flows, provided that the swirl level is not too high and the measurement volume sufficiently far away from the probe tip. Sensitivity to swirl is not expected to be significant when performing measurement in swirling flames, since swirl-induced effects are substantially reduced if compared to corresponding cold flows with identical inlet swirl level. For example, the flow pattern of a swirling flame with a 0.6 inlet swirl number will exhibit similarities to a cold flow of 0.15 to 0.2 inlet swirl. This was attributed to an increase of the axial momentum with respect to the tangential momentum when combustion takes place (Weber & Ougue, 1992). Thus, the probe intrusiveness is not expected to be a problem in swirl flames unless the inlet swirl level is extremely high. RESULTS The following sections present examples of laser velocirnetry measurements in natural gas, heavy fuel oil and coal flames. Complete reporting of all associated measurements (temperature, gas composition) and analyses is available in [14-16]. Measurements in a 12 MW Natural Gas Flame These measurements were carried out by an IFRF staff at the JOHN ZINC company, Oklahoma, in the test furnace n03 as part of an investigation on the scaling of natural gas burners [14]. The furnace has an internal diameter of 3.35 m and a length of 9 m. Standard probes were used to measure gas temperature and composition (02, CO, C02, NOx, and unburned hydrocarbons). LOV measurements were performed at six measurement stations in three 12 MW flames. The flow was seeded with about 10 kg/h of Zirconium oxide powder classified in a size range of 0.5 to 5 Jim. The seeding powder was injected in the combustion air stream with a scale-up version of a cyclone seeder described in [17]. Figure 8 displays the axial and tangential velocity profiles for a flame with an inlet swirl number of 0.6. The inlet swirl number was obtained by integrating the axial and tangential velocity profiles measured in the inlet duct, between the swirl generator and the quart inlet, as described in [5, 13J. The top half of the furnace displays the mean and RMS axial velocities (U and u'), while the bottom half displays the mean and RMS tangential velocities (Wand w'). For readability reasons, the velocity scale factor at the first traverse is twice larger than for the next five traverses. The mean axial velocity (U) profiles display a narrow but strong internal recirculation zone just downstream of the burner centerline, with negative velocities as high as -12 m/s. As shown in previous studies on swirling flames [5, 13J, the central recirculation zone results from using a natural gas injector with a large burner blockage ratio and radial gas injection. The reverse flow zone is further promoted by the presence of a divergent quarl. Further downstream, the profiles show that the axial velocity maxima converge to the centerline and that the flow evolves rapidly into a constant tangential velocity profile. 1 em = 15 mls o 2.54 m Figure 8: LDV measurements in a 12 MW natural gas swirl flame Measurements in a 2.5 MW Coal Flame These measurements were carried out in the IFRF Furnace no 1. The furnace has an internal diameter of 2 m and a length of 6 m. These experiments were part of series of measurements aimed at generating a complete set of experimental data for mathematical model validation [15]. Standard probes were used to measure gas temperature and composition (Ch, CO, CCh, NOx, HCN and NH3). Figure 9 displays the axial and tangential velocity profiles at the four measurement stations. LOV measurements performed without artificial seeding resulted in data rates in the range of a few kilohertz. The high data rate was attributed to the high number density of coal particles covering a large size range. The application of laser velocimetry in coal flames requires assessing the effect of particle slip on the velocity measurements. In regions close to the fuel injector, the small particles (up to 10 Jim) are expected to follow the gas phase, whereas the large particles (75 to 200~) will retain their original momentum over a longer distance and will be less influenced by the gas phase turbulent fluctuations. The Mie theory indicates that for particles larger than the illuminating light wavelength, the scattered light intensity increases with particle size. It should therefore be possible to use the burst pedesdal amplitude to obtain some indication on the significance of particle slip. Howe~er, the application of this technique is limited by a serious of shortcoming: large particles produce low amplitude signals when passing through the edge of the measurement volume, and are thus identified as small particles. As a result, the signal amplitude cannot accurately separate signals from small and large particles and can at best provide only a qualitative indication of the presence and direction of particle slip. A safe stand is to use the amplitude/velocity correlation to assess only whether particle |