| dc.description.abstract | Distributed roughness occurs over a variety of aerodynamic surfaces such as wind-/gas-turbine blades and aircraft wings in the form of dust, insect strikes, fouling or icing. The disturbances introduced by the roughness cause early boundary layer transition, which manifests as reduced range in aircraft and lower power production from the turbines. The theme of the present thesis is to characterize and control the boundary layer transition caused by distributed roughness. In this context, we address the following problems
• Characterizing the pre-transitional region of the boundary layer subjected to distributed roughness
• Investigating the mechanism of boundary layer transition caused by distributed roughness
• Using two-point correlation and proper-orthogonal-decomposition (POD) analyses to visualize the wall-normal spread of the disturbances in bypass transition and to quantify the penetration of freestream disturbances
• Delaying boundary layer transition caused by distributed roughness and controlling extreme events in transition induced by freestream turbulence (FST)
Towards this, we carry out wind tunnel measurements over a flat plate using sandpaper strips to simulate the effect of distributed roughness. Single-component hot-wire anemometry and two-dimensional particle image velocimetry (PIV) measurements are used to characterize bypass transition caused by various disturbance sources. A coarse 24-grit roughness strip is used as the primary disturbance source for most of the measurements. Additional measurements are conducted for other types of disturbance sources as well, such the 80-grit roughness, a combination of different roughness strips and a turbulence grid. We also present new analyses on previously acquired in-house PIV data with an isolated roughness/turbulence grid as the disturbance source. Characterization of the 24-grit sandpaper strip using a laser scanner shows the presence of roughness grits of different heights that are nearly randomly distributed over a backing cloth indicating that a wide range of length scales are introduced by the roughness into the boundary layer. A careful error analysis is carried out to determine the uncertainty levels for the hotwire and PIV measurements.
The pre-transitional regions downstream of the 24-grit and 80-grit roughness are characterized using hotwire anemometry. Using spectral analysis of the hot-wire signals, we show that the roughness strips introduce a broad range of disturbances into the boundary layer along with spanwise vortices shed from the step created by the roughness. We demonstrate that the “sub-optimal” nature of disturbances introduced by the distributed roughness results in unfavorable comparisons with theoretical calculations based on optimal initial conditions, unlike freestream turbulence which is known to introduce “optimal disturbances”. The analysis also shows the difference between the disturbances introduced by the two roughness strips with the 80-grit roughness demonstrating a linear growth of disturbance energy consistent with transient growth theory, whereas the results for 24-grit roughness deviate substantially from theoretical predictions.
A detailed characterization of the boundary layer downstream of the 24-grit distributed roughness is conducted using PIV. Results show that distributed roughness causes steady streaks with strong spanwise inhomogeneity, even though the sandpaper strip consists of roughness particles that are nearly randomly distributed. Consistent with recent numerical studies, we find that the transition mechanism involves unsteadiness developing locally on the strongest streaks, indicating that the spanwise organization of streaks is an important factor in predicting the onset of transition. Further, we demonstrate that the transition mechanism downstream of the distributed roughness involves both “inner” and “outer” modes of instability similar to FST-induced transition, which is in contrast to the transition downstream of an isolated roughness, where only one instability mode is dominant. Analysis of the PIV visualizations and conditional sampling of velocity data show that after the onset of transition, there are several qualitative similarities between the different disturbance sources (FST, isolated and distributed roughness) vis-à-vis instability structures, and distribution of streaks, which can be helpful in modelling bypass transition. We discuss the difficulty involved in comparing different transition scenarios due to the wide parameter range associated with each disturbance source and clearly bring out the scope of the present comparison in the context of the parameter values used in this study.
The role of unsteady streaks in the transition mechanism for distributed roughness is further explored by POD analysis, which shows that the dominant modes (with highest energy) in the pre-transitional and transitional regions contain streaky structures. We use a recently proposed technique of using POD modes having large coefficient values to reconstruct characteristic transitional features for the distributed roughness transition dataset. We demonstrate that for individual instances, selecting POD modes with large coefficients results in a better reconstruction of the flow field, for a given number of modes, in comparison to the traditional method of choosing modes having large energy. A threshold for the coefficient values that can reconstruct transitional characteristics satisfactorily is identified, which is expected to be helpful in developing low-dimensional models for bypass transition. Next, we investigate the utility of the two-point fluctuating velocity correlations in quantifying the spread of disturbances in the wall-normal direction and demonstrate that the disturbances introduced by the distributed roughness extend well into the freestream. This technique is further extended to visualize the penetration of freestream disturbances into the boundary layer in FST-induced transition. We propose a metric for quantifying the penetration depth using the integrated value of velocity cross-correlation (between wall-normal and streamwise velocity components) in the boundary layer and show that this metric is more robust in comparison to the existing techniques. We also make a direct comparison between the estimates of penetration depth obtained from the present method with the available theoretical results in the literature and find a satisfactory agreement.
The improved understanding of the distributed roughness transition is utilized to develop a passive method for mitigating the effect of naturally occurring (primary) surface roughness in aerodynamic devices using fine (secondary) roughness strips placed upstream and/or downstream. A combination of roughness strips placed upstream and downstream of the primary roughness is demonstrated to result in maximum transition delay. Smooth strips placed upstream and downstream are also shown to be effective in causing transition delay. A parametric study on the length of the downstream secondary roughness indicates that there is an optimum length of the downstream roughness that results in maximum transition delay and increasing the length of secondary roughness beyond this does not improve aerodynamic performance. Analysis of the velocity signals downstream of the different configurations indicate that the spanwise vortices shed from the roughness play a key role in the transition process. The downstream roughness reduces the strength of spanwise vortices shed from the primary roughness whereas upstream roughness lifts the boundary layer, resulting in a lower effective roughness Reynolds number. Using spectral and statistical analysis of fluctuating velocity data in the transitional region, we demonstrate that the secondary roughness strips do not alter the statistical structure of the boundary layer after the onset of transition, making the present technique well suited for applications where transition delay is desired without significantly altering the transition zone characteristics. A passive, lightweight technological translation of this work is proposed, applicable to wind turbine blades and aircraft wings. A patent has been filed on this invention to the Indian Patent Office.
The idea of using secondary roughness strips is extended to FST-transition, where we demonstrate using hotwire measurements that some transitional features can be controlled using surface roughness. We find that strategically positioned roughness strips can alter the spectral structure of the boundary layer in a way that reduces the maximum level of fluctuations and instances of high amplitude streaks. The roughness placed in pre-transitional region is demonstrated to be more effective in this respect. We expect this to be of use in turbomachinery flows, where secondary roughness can be used to reduce the number of “hot streaks” in the transitional region that can cause thermal damage to turbomachinery blades. | en_US |
