1 March 2022 in Physics of Fluids
Direct numerical simulations (DNSs) of an incompressible turbulent boundary layer on an airfoil (suction side) and that on a flat plate are compared to characterize the non-equilibrium turbulence and the effect of wall curvature on the flow. The two simulations effectively impose matching streamwise distributions of adverse pressure gradient (APG) quantified by the acceleration parameter (K). For the airfoil flow, an existing compressible DNS carried out by Wu et al. [“Effects of pressure gradient on the evolution of velocity-gradient tensor invariant dynamics on a controlled-diffusion aerofoil at Rec = 150,000,” J. Fluid Mech. 868, 584–610 (2019)] of the flow around a controlled-diffusion airfoil is used. For the flat-plate flow, a separate simulation is carried out with the aim to reproduce the flow in the region of the airfoil boundary layer with zero to adverse pressure gradients. Comparison between the two cases extracts the effect of a mild convex wall curvature on velocity and wall-pressure statistics in the presence of APG. In the majority part of the boundary layer development, curvature effect on the flow is masked by that of the APG, except for the region with weak pressure gradients or a thick boundary layer where the effect of wall curvature appears to interact with that of APG. High-frequency wall-pressure fluctuations are also augmented by the wall curvature. Overall, the boundary layers are qualitatively similar with and without the wall curvature. This indicates that a flat-plate boundary layer DNS may serve as a low-cost surrogate of a boundary layer over the airfoil or other objects with mild curvatures to capture important flow features to aid modeling efforts.
31 March 2023 in Journal of Fluid Mechanics
Wall-pressure and velocity statistics in the turbulent boundary layer (TBL) on a cambered controlled-diffusion aerofoil at $8^{\circ }$ incidence, a Mach number of 0.25 and a chord-based Reynolds number ${Re}_c=1.5\times 10^{5}$ are analysed at four locations on the suction side with zero and adverse pressure gradients (ZPG and APG), characterised by increasing Reynolds numbers based on momentum thickness, ${Re}_{\theta }=319$ , 390, 877 and $1036$ . The strong APG yields a highly non-equilibrium TBL at the trailing edge that significantly affects the turbulent flow statistics. Different normalisations of the full wall-pressure statistics involved in trailing-edge noise are analysed for the first time in such strong APG with convex curvature, and compared with available experimental and numerical data. Good overall agreement is found in the ZPG region, and most results obtained in previous APG TBL can be extended to the present highly non-equilibrium case. The presence of strong APG augments the intensity of wall-pressure fluctuations noticeably at low frequencies, shortens the streamwise and broadens the spanwise coherence of wall-pressure fluctuations in both time and space, and significantly reduces the convection velocity. The wall-pressure power spectral density are found to scale with the displacement thickness, the Zaragola–Smits velocity and the root-mean-squared pressure, the latter possibly being replaced by the local maximum Reynolds shear stress. The other two key parameters to trailing-edge noise modelling, the spanwise coherence length and the convection velocity, rather scale with displacement thickness and friction velocity, respectively.
16 February 2021 in Journal of Turbulence
This paper explores the use of a small-span direct numerical simulation for a transient, smooth-wall turbulent channel flow and then applies the small-span simulation to a transient channel flow with riblets. A flow configuration similar to that of S. He and M. Seddighi (J Fluid Mech. 2013;715:60–102) is used to study the impulse response of a half-height channel flow to an abrupt increase in bulk velocity (with a friction Reynolds number increasing from 180 to 418). A minimal domain span sufficient to include the near-wall quasi-streamwise vortices in the ‘healthy turbulence’ region is used. The turbulent flow undergoes reverse transition toward a quasi-laminar state, followed by a retransition phase to the new equilibrium state. On a smooth wall, detailed comparisons with a full-span case show that the small-span test case captures satisfactorily the essential dynamics during the entire transition process, although it yields a slight delay in recovery to the new equilibrium. This difference is attributed to a slower streak transient growth due to an underestimation of near-wall spanwise fluctuations. This underestimation is associated with the missing large attached eddies that are not contained in the small span of the simulation domain. These comparisons justify the use of small-span simulations for identifying the main flow physics in a non-equilibrium accelerating wall turbulence. The application to the riblet flow shows that riblets do not fundamentally affect the flow dynamics, but delay the retransition as a result of significantly milder streak meandering. The streak-stabilisation effect of riblets is still active in a strongly accelerating turbulence and tends to prolong the flow recovery.
1 May 2025 in Noise Control Engineering Journal
This paper investigates the application of machine learning regression algorithms—Kernel Ridge Regression (KRR), Huber Regressor (HR), and Gaussian Process Regression (GPR) for predicting sound power levels of gensets, offering significant value for marketing and sales teams during the early bidding process. When engine sizes and genset enclosure dimensions are tentative, and measured noise data is unavailable, these algorithms enable reliable noise level estimation for unbuilt gensets. The study utilizes high-fidelity datasets from over 100 experiments conducted at Cummins Acoustics Technology Center (ATC) in a hemi-anechoic chamber, adhering to ISO 3744 standards. By using readily available information from the bidding and initial design stages, KRR predicts sound power with an average accuracy of ±5 dBA. While HR and GPR show slightly higher prediction errors, all models effectively capture the overall noise trends across various genset configurations. These findings present a promising method for early-stage noise estimation in genset design.