3.1 Data Validation

Each data set is identified first by a seal flow rate setting, second at a vane position and third at a rotor position. Before close analyses of the data are performed, the integrity of the data set is investigated by comparing various measures of mean velocities. Consider three different types of velocity measurements, represented in four forms. Turbine flows were measured via PIV interrogation of a specific area, Pitot probe measurement of the inlet midspan velocity, and venturi metering of the mass flow through the machine. A comparison is then made between individual PIV data planes, between groups of PIV data planes, and between the three flow measurements.

First consider a comparison between the three flow measurement types. The calculated average of all axial components of all the valid vector information from PIV results in a mean velocity of 88.6 ft/sec (27 m/s). This matches well with the midspan Pitot probe measured axial inlet velocity of 93.5 ft/sec (28.5 m/s), and the annulus area averaged inlet velocity from mass flow measurement which was 94.2 ft/sec (28.7 m/s). From this conclusion it can only be verified that the PIV measurements are of a very reasonable magnitude. A more direct comparison cannot be easily made between these inlet axial velocities and the PIV measurement cases, since the vector information is not measured at the vane inlet and includes only the slowest 33% of the span.

In addition to acceptable comparison of the axial inlet velocity to the PIV measured data, plane to plane comparisons, rotor position comparisons and seal flow rate comparisons can be made of the PIV data. Figure 3.1 shows data plane average axial velocity values for each rotor position and vane position at the high and low seal flow rates, where each plot point is the average of all valid vector axial components for that data plane. The graph abscissa denotes rotor position according to the data acquisition delay time for the measurement plane. The four vane positions are marked by line styles and labeled as Plane A, B, C and D. Two seal flow rates are represented in Figure 3.1, with a dark line indicating the high seal flow rate and a light gray line marking the low seal flow rate.

Figure 3.2 shows the average radial velocity for each data plane in the same way that Figure 3.1 shows average axial velocity. In analyzing these data, it was noted that for the case of the low flow rate, at measurement Plane D, plot peaks occur at times 375 ^s and 525 ^s. It would be expected that these values should align with the Plane D high flow rate peaks at time 450 ^s and 600 ^s. The discrepancy is more noticeable in the axial velocity plot, where Plane D data for the low flow rate lags behind Planes A, B and C. The conclusion that this data set is out of phase with the other data at the indicated blade pass times is also supported from analysis of corresponding vorticity plots and vector plots [Appendix A,B,C and D: compare Figure B.5 and Figure B.6 vs. Figure A.6 and A.7 or Figure D.5 and Figure D.6 vs. Figure C.6 and Figure C.7]. Thus, the data in question is marked with a "P" (partial set) in Table 3.1 and care must be taken in drawing conclusions from this data set (low seal flow rate, Plane D). For the purpose of studying circumferential flow variations, the high flow rate case must be used if the full range over which data was acquired is to be included.

Other than the peak lag in Plane D, vane plane to plane variations in radial averages are insignificant considering the effect of secondary flow on the radial component of velocity within the measurement region. However, the variation in average axial velocity with vane position is significant, with a total range of over 65 ft/sec (20 m/s) from vane Plane A to vane Plane D, and at a fixed rotor position the variation is still as much as 40 ft/sec (12 m/s). These differences are likely a result of turning variation within a vane passage, a hypothesis that is consistent with increased axial velocities at positions further from the vane surface.

The oscillation in average velocity for rotor position is a result of the rotor blade flow blockage. As the blade passes the measurement plane, the flow decelerates axially, and turns around the approaching flow obstruction. This event leads the rotor potential field maximum event which occurs at time t=675^s. The fact that the rotor leading edge pass event lags the time of minimum average axial velocity can be rationalized by understanding that the measurement plane is skewed 50° from the rotor relative inlet flow angle thus the streamline direction turns toward the tangential direction prior to rotor pass and then back towards axial after rotor pass as it accelerates around the blade edge. Continuity arguments then explain the variation in axial velocity around the rotor pass event.

Trends in average axial velocity between two different seal flow rate cases are not as easily explained. The observed changes are likely due to the turbine set point, since duplicating the flow coefficient and loading coefficient setting when changing seal flow rate required a slightly different set of blower damper and load control settings. There is thus a small variation in turbine operating point from case to case. Overall the data sets presented are determined to be acceptable for meaningful comparisons of flow development in the seal region.

3.1.1 Ensemble Averaged Data

Each data set presented in the Table 3.1 is a result of an average performed on a collection of approximately 100 instantaneous measurements at each vane position, phase locked to a single rotor position. This ensemble averaging process serves to remove aperiodic unsteadiness that exists within a turbulent flow, while preserving all harmonics of the rotor pass event. Thus, the effects of the first stage rotor blade potential field are extracted from background unsteadiness. The degree of unsteadiness in the flow field is quantified in Figure 3.3, where a weighting scheme is presented to show the standard deviation of the axial velocity components within each interrogation area. The standard deviation of the 100 instantaneous measurements is a measure of the uncertainty of the resulting average value which is due to the flow unsteadiness.

Table 3.2 presents the key to this weighting scheme, which also includes a consideration of the quantity of vectors from each of the 100 instantaneous measurement points that were validated through the PIV software. A vector is considered valid according to the online PIV processing if it was within a magnitude range limit of -65.6

to +196.85 ft/sec to (-20 to +60 m/s) and the cross-correlation primary peak to secondary peak ratio is greater than Rp=1.1, which is a measure of the signal-to-noise ratio. Previous PIV investigations (Keane and Adrian, 1992) have shown that Rp=1.2 is a good limit for the peak ratio, however low seed density, uneven lighting and strong secondary flows require a less restrictive limit for minimizing loss of data.

In the presence of a vane wake, and with a strong rotor potential field acting on the measurement region, Figure 3.3 shows that rather wide velocity distributions exist within the unsteady flow. Although the highest standard deviations are greater than 63% of the overall average axial velocity, discarding data on this basis could remove vortical structures of interest from the processed vector field. Thus ensemble averaged velocity vectors were included in the results analysis as long as greater than 40 valid instantaneous vectors existed at the particular interrogation region (weighting scale level 8).

Table 3.2 Axial Velocity Weighting Scale for Figure 3.3.

Weighting Scale

Scale Definition

0 0

Post a comment