II. LA EXPERIENCIA AMAZÓNICA
2.1. ANTECEDENTES Y CONTEXTO
are shown in Figure 5.26 - Figure 5.29 and were computed at G = 1.93 kg/sm2bar. These predicted results were averaged over the X2 surface area cooled by each hole and the
experimental results only surface average by conduction in the metal wall. The thermocouple position in the lowest heat transfer region will underestimate the true local surface averaged h due to the residual temperature gradients in the wall, as the Biot number was low but not 0, as Figure 5.30 shows for varied Z/D. The extent of these temperature gradients was also computed and will be shown later.
Figure 5.26 (a and b) shows the prediction of the surface averaged h (over X2) as a function of the number of holes in the cross-flow direction for the range of Z/D values. The predictions are shown to agree well with the experimental results at Z/D from 1.22 - 6.42, which are all within 5% error as shown above. But for Z/D of 0.76 in Figure 5.26 (a), the agreement was poor and this is because of the pressure loss disagreement, shown in Figure 5.12 caused by highly reduced grid in the impingement gap.
(a) Smaller range of Z/D at fixed G of 1.93 kg/sm2bar
(b) Larger range of Z/D at fixed G of 1.93 kg/sm2bar
(a) Higher X/D at fixed n
(b) Lower X/D at fixed n
Figure 5.27: Comparison of X/D target X2 average HTC h for constant G of 1.93kg/sm2bar Figure 5.27 (a) shows the locally X2 surface average HTC h predicted and experimental results for the range of higher X/D of Table 4.12. The experimental results shows that for X/D of 11.04, the heat transfer was fairly uniform with a slight increase in the trailing edge region, possible due to the duct flow additional heat transfer of the cross-flow. These results indicate that at an X/D of 11.04 the deterioration of heat transfer with distance, as correlated by Equation 2.22, does not occur. This supports the prediction of the aerodynamics shown by El-jummah et al [25], where there was minimal movement of the jets by the cross-flow. The agreement of the CFD results with the experiments was rather poor for X/D of 11.04, apart from in the leading edge region. The reason for this might be associated with the use of incompressible flow CFD, when at this X/D the jet velocities are very high at 244 m/s, as Table 4.12 shows, where compressible flow CFD should be used.
For X/D of 6.54 and 4.66, Figure 5.27 (a) shows that there was very good agreement between the experimental and the CHT CFD results. Both the experiments and predictions showed the deterioration of h with the cross-flow, due to the downstream convection of the turbulence on the surface, as shown above and correlated in Equation 2.22. This is in a region of X/D where flow-maldistribution was not significant. However, at lower X/D of Figure 5.27 (b), the experimental results showed first a decrease in h with distance along the impingement gap, due to the cross-flow effect correlated by Equation 2.22 and then an increase due to the influence of flow-maldistribution. This effect was reasonably well predicted using CHT CFD for an X/D of 3.76, although the leading and trailing edges were under predicted and the central section over predicted. Figure 5.27 (b) shows that the difference in the predictions and measurement were highest at the leading edge.
Finally, at the lowest X/D of 1.86 the experimental and predicted results were in agreement over a continuous increase in h from the start to end of the gap, due to the strong flow- maldistribution at this X/D with a very low impingement jet pressure loss. However, the predictions under predicted the experimental results at all axial positions. The difference was 50% at the start of the gap and 18% by hole 10. As discussed above in relation to the surface averaged h results, these results are difficult to explain as predicted h is higher than that measured would be expected, due to the location of the thermocouples on the centreline between the impingement holes. The prediction of the X/D of 1.86 in Figure 5.6 showed very low TKE over the first few rows of holes, which was due to the very low predicted proportion of flow in these holes as shown in the very low predicted velocities by El- jummah et al [25]. The resultant predicted flow-maldistribution in Figure 5.3 was severe with > 5 times the flow in row 10 to that in row 1. Table 4.12 gives the hole Re as 3850, based on the assumption of equal distribution of the air flow. However, the Re in the first row of holes with the predicted flow-maldistribution was 1540 and the applicability of the turbulent flow modelling under very low jet Re conditions is probably the main cause of the prediction errors for the X/D = 1.86 impingement geometry.
The influence of cross-flow on the average surface heat transfer per hole, using the surface area X2 for each hole is shown in Figure 5.28 for range of G from 10.35 - 1.93 kg/sm2bar. These are the average of the surface distributions of heat transfer shown in Figure 5.17 (id and iid) for G = 1.93 kg/sm2bar and the equivalent for other decreased G [46]. Figure 5.28 shows very good agreement between the experiments and the predictions at the highest G values of 1.93kg/sm2bar, with a trend similar to the highest G prediction for G = 1.48kg/sm2bar. The agreement is also good at G of 1.08kg/s.m2bar, except at the leading edge two holes where the predictions were significantly below the measurements. El- jummah et al [46] showed this should be based on lower Nu for the three leading holes and
the absence of a cool spot in the predicted temperatures. A possible reason for this could be an over prediction of the flow-maldistribution, so that the leading edge holes were predicted to have a lower velocity than actually occurred. Figure 5.4 shows some unexpected variations in the flow mal-distribution, which should have been a smooth variation with the number of holes.
Figure 5.29 shows the predicted X2 surface average HTC for varied n at fixed X/D of 4.66, as a function of the distance along the impingement cross-flow gap. The predictions are expected to be slightly higher than the measurements as the thermocouples on which the measurements were based were located at the lowest local heat transfer position and relied on the metal wall thermal conduction to produce the surface averaging. This was done so that the experimental results, which were used directly in combustor wall cooling designs, would be conservative.
Figure 5.28: Comparison of target X2 average HTC h for range of G at fixed X/D and Z/D
Figure 5.30: Target wall X2 average Biot number for range of Z/D at fixed X/D and G For n = 26910 m-2 the predicted local h were all slightly higher than the measurements, as expected, as the thermocouple was averaging the heat transfer of more than one impingement jet. The predictions for n = 1076 m-2 were higher than the measurements, this was because the thermal gradients were highest at this wide jet spacing and the thermocouple was furthest from the impingement point. The agreement of the predictions and measurements was very good for n = 4306m-2. For n = 9688m-2 (15 × 15) the predictions were quite different from the measurements as the predictions were significantly below the measurements at the leading edge of the impingement gap and higher at the trailing edge. This resulted in good agreement for the mean surface average h predictions and measurements, as shown in Figure 5.25.