OCR Text |
Show measure the gas flow velocity. T w o laser beams are focused at a common crossing point in the gas flow. At the crossing point, the reinforcement and cancellation of the light waves produces an interference fringe pattern within the focal volume.10 W h e n a particle passes through the focal volume, the intensity of the laser light scattered by the particle will be modulated by the light and dark interference fringes. T he scattered light's amplitude is monitored as a function of time by a high frequency response photomultiplier tube. The frequency of the modulation is equal to the particle velocity times the interference fringe spacing. A digital signal processor analyzes the photomultiplier signal, identifies and validates the Doppler bursts, extracts the modulation frequency information, and calculates the velocity of the particle. T o tell the direction of motion a known frequency shift is applied to one beam. This makes the fringe pattern move in space, and allows one to distinguish between positive and negative particle velocities (up to the amount of the frequency shift). The focusing lens was mounted on a linear translation mechanism that is equipped with a lead screw and stepper motor so that the position of the beam crossing could be moved. This ability to translate the focusing lens and thus the beam crossing point enabled us to move the measurement volume within the furnace and measure the velocity at various distances from the furnace wall. In order to measure the velocity profile from near the wall to the maximum distance allowed by our longest focal length lens, we used three different lenses (75-cm, 120-cm, and 150- c m focal lengths). By using three different focusing lenses, the measurement distances overlapped, thus allowing for full coverage. The combustion gases inside this gas-fired furnace were very clean with very few particulates. Therefore, we had to supply some particles so that the L D V system would work. This was accomplished by inserting a water-cooled, stainless steel particle injection probe into the furnace. This probe had an outer diameter of -16 m m (5/8") and could be inserted up to 1.5 m (5 ft) into the furnace. For seeding, we used alumina (aluminum oxide) particles with a nominal diameter of 20 pm. Alumina was chosen because it could withstand the high temperature found in the furnace. The alumina particles were placed in an Erlenmeyer flask that is closed with a two-holed stopper. A flow of nitrogen (about 315 cm3/sec (40 scfh))was introduced through one hole, causing the alumina particles to become entrained in the nitrogen. The nitrogen and particles then flowed out the other hole in the stopper, through some tubing, and then through the hollow probe into the furnace. The insertion depth was adjusted to inject these particles near the L D V measurement volume. The contents of the flask were stirred with a magnetic stir bar to keep the alumina particles flowing smoothly. Typically w e would fill the flask with 100 to 200 grams of alumina powder, and this would last for roughly an hour. Thus the total amount of alumina and nitrogen added via our probe was extremely small compared to the size and flow rates of the furnace. The results obtained through one of the furnace ports is shown in Figure 4. The vertical component of the gas velocity is shown as a function of distance from the outside of the furnace wall. W e define the upward direction as positive so the positive velocity indicates that the furnace gases are flowing upward along the wall. Different symbols are used to designate the data collected with the three focal length lenses so that the agreement of the different measurements can be appreciated. The first three data points correspond to measurements from inside the viewport hole. The quality of the data sets varies significantly. Certain data sets, collected when the measurement volume was positioned close to the wall, contain hundreds of valid Doppler bursts, enabling determination of the velocity very precisely. Other data sets, typically those collected far into the depths of the furnace, contain only a few valid Doppler bursts; consequently the error bars for these data sets are much larger. The error bars shown in Fig. 4 are the 9 5 % confidence interval for the mean, computed by dividing the standard deviation of the data set by the square root of the number of data points and then multiplying by 1.96. W e experimented with robust statistics to improve our results, especially far from the furnace wall where the data rate was low and oudiers were more prominent. The median is a more robust measure of central tendency than is the mean, and the median absolute deviation is a more robust measure of spread than is the standard deviation. The use of medians and median absolute deviations, in general, reduces the uncertainty, but the overall pattern of Fig. 4 remains: once inside the wall, the vertical component of gas velocity is nearly constant at a few m/sec. CONCLUSIONS The selected results presented here were collected by DIAL during a week-long, on-site field measurement campaign to characterize an industrial process furnace. They illustrate the types of results that can n o w routinely be -9- |