EXPERIMENTAL STUDY OF REDEVELOPING BOUNDARY LAYER FLOW PAST FENCES

Abstract

It is quite common in nature that a turbulent boundary layer is disturbed due to change in the surface terrain or due to a surface irregularity. In the latter case the boundary layer separates, reattaches and then redevelops. Atmospheric boundary layer flow past multiple buildings or a number of rows of shelter belts are a few examples of such flows. The flow past solid and porous sharp-edged fences, one placed on the downstream side of the other in a turbulent boundary layer, have been studied in this investigation to provide information regarding form drag of the fences, mean velocity distribution and distribution of longitudinal intensity of turbulence. Experiments were conducted at zero pressure gradient in a closed circuit wind tunnel with a test section of 0.81 m x 1.14 m and 3.05 m long. The Reynolds number (U^H/i/) was varied from lxlO4 to 6.9 x 104. Two-dimesional fences of height varying from 10 mm to 40 mm and porosity from 0 to 0.38 were used. The variation of H/5q was from 0.422 to 1.688, H/h from 0.6 to 4and X /H from 8 to 32. (Here H is the height of the main fence, h is the height of secondary fence, ^ is the distance of the secondary fence from the main fence and 5q is the nominal boundary layer thickness of the undisturbed boundary layer) The drag on the fences was measured with the help of a force transducer and the mean and turbulence velocity profiles were obtained with the help of a Pitot tube - Baratron assembly and a constant temperature hot wire anemometer respectively. ii From the analysis of data on drag of a fence placed downstream of another fence, it was concluded that the outer region of the velocity profile influences the drag on the fence significantly. This is contrary to the implication from the relationship of Ranga Raju et al. (106) that only the velocity profile in the inner region is important in determining the drag of fences. The analysis of the data also indicated that for all the data for undisturbed and disturbed boundary layer flow past solid and porous fences there is a unique relationship between C (1+ti1-5) and P(9/5.)0-5. This relationship is valid for two D2 dimensional solid and porous fences placed in a flow with zero pressure gradient for the cases of disturbed and undisturbed boundary layer developed over smooth, transitional and rough surfaces. Two models were proposed for the analysis of the mean velocity profiles in the flow past two fences. (i) Model I:- The secondary fence is assumed to be removed and the mean velocity profile downstream of the main fence predicted as per Gupta and Ranga Raju (52 and 53). The difference between the observed and predicted mean velocity (Au) can be ascribed to the presence of the secondary fence and analysed further, (ii) Model II:- This model is based on the premise that the secondary fence may be presumed as being placed in a thickened boundary layer developed due to the main fence. The velocity profiles downstream of the secondary fence placed in such a boundary layer can again be predicted using the method of Gupta and Ranga Raju (52, 53). The difference between the observed and iii the predicted mean velocity profile (Au) may be analysed further as the effect of interference of fences. In the flow past two fences, two regions of flow may be> identified: (i) Region between the fences (ii) Region beyond the secondary fence. _ The analysis of the mean velocity profiles has shown that Model I is appropriate for the region between the fences while Model II has been found to be more suitable in the region beyond the secondary fence. Predictors for Au have been developed in both the cases for mean velocity in case of solid fences. It has, however, not been possible to develop predictors for mean velocity in case of porous fences and more data are required to evolve predictors in this case. The data on longitudinal turbulence intensity collected in this study have first been used to modify the relationship for J^2 proposed by Ranga Raju et al. (105) for the case of a single fence, since such modification was indicated to be justified by the data. These modified relations have then been used as the basis to compute ftfiu72] in the appropriate model for the data and then develop predictors for these in the same manner as for mean velocity profiles. The agreement of the experimetal data with the profiles predicted using the corrections 'Au' for mean velocity predictions for solid fences and ^^- X100 for longitudinal intensity of turbulence for both solid and porous fences has generally been found to be good.