Analysis of flow structures in cavity flow with modifications to cavity geometry

• Main Author: Pey, Yin Yin
• Other Authors: Chua Leok Poh
• Format: Theses and Dissertations
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering::Aeronautical engineering; DRNTU::Engineering::Mechanical engineering; DRNTU::Engineering::Mechanical engineering::Fluid mechanics; DRNTU::Engineering::Aeronautical engineering::Aerodynamics
• Online Access: https://hdl.handle.net/10356/59862
• Institution: Nanyang Technological University

Flow over cavity exists in a wide range of flow phenomena and remains a challenging problem till today. The objective of this study is to investigate the physics of cavity flow in response to trailing edge modifications. Trailing edge ramps were selected as the passive control method to be studied as they have the potential to reduce high pressure impacts caused by the impingement of shear layer. Investigations were carried out using extensive steady pressure measurements and Particle Image Velocimetry (PIV) at a freestream velocity of 15m/s. Six different combinations of ramp heights (hramp/D = 0.25 and 0.5) and ramp angles (θramp = 30o, 45o and 60o) were examined individually for pressure drag reduction in baseline cavities of length to depth ratio (L/D) 2, 3, 3.5, 4, 6 and 8. In particular, an in depth study using PIV was done for open cavities of L/D = 2, 3 and 3.5. This gave insights into the flow behaviours upon ramp implementations. From the extensive pressure measurements, it was found that for all L/D tested, a θramp = 30o resulted in the lowest estimated pressure drag. For L/D = 2, this occurred when hramp/D = 0.25. As for L/D = 3, 3.5, 4, 6 and 8, this occurred when hramp/D = 0.5. The crucial angle of 30o appears to minimize the shear layer impingement on the trailing walls by allowing the flow to move smoothly out of the cavity. PIV measurements complemented the pressure measurements by providing information on the Reynolds stress distribution within the cavity, hence revealing regions of quieter flows in the presence of trailing edge ramps. To exemplify the redistribution of energetic structures within the cavity in the presence of trailing edge ramps, the technique of Proper Orthogonal Decomposition (POD) was employed. A comparison of the modified cases (θramp = 30o, hramp/D = 0.5) against the respective baseline cases for L/D = 2 and 3 shows a shift in the energetic structures from within the cavity in the baseline cases to occupy the shear layer in the modified cases. This resulted in a quieter flow within the cavity where the energetic structure used to exist. Based on the more energetic modes of flow, a simplified reduced order model aimed at reconstructing the time evolution of flow was attempted. This was done using the first three spatial modes and an estimation of the temporal coefficients using a sinusoidal function. The reduced order model was tested on a modified cavity of L/D = 3, θramp = 30o and hramp/D = 0.5. The predicted flow revealed high accuracy in recovering the general flow features of the cavity when compared against the original instantaneous snapshots. Another method of flow prediction using Stochastic Estimation (SE) was looked into. Both the linear and quadratic estimations were performed on a baseline case of L/D = 3 and its modified case with θramp = 30o and hramp/D = 0.5. The estimator used was velocity magnitude while the estimated quantities were velocity fluctuations in the streamwise and vertical direction. The advantage of this method lies in using only PIV data to perform the estimation. The linear and quadratic estimations predicted the baseline cavity flow to the same extent while the quadratic estimation proved more superior than the linear estimation in predicting the flow of the modified case. In both test cases, velocity fluctuations in the streamwise direction were predicted rather accurately while velocity fluctuations in the vertical direction were under-predicted. This was found to be attributed to the poor correlation between the estimator and estimated quantity. The significance of these results shows how trailing edge ramp implementations can reduce pressure drag and the configurations that are optimal in each L/D. The change in flow behaviours that bring about this decrease in pressure drag was also presented. Finally, different methods of predicting and reconstructing the instantaneous flow were also examined.