Optimal Control of Turbulent Channel Flow Using an Les Reduced-Order Model (original) (raw)
Related papers
Active control of turbulent channel flow using Large Eddy Simulations
In this study, we are interested in reducing the drag force on the upper and lower walls of a plane turbulent channel flow, using an active flow control scheme. Large eddy simulations (LES) and optimal control theory are used to determine control laws that effectively reduce the turbulent kinetic energy and drag of a turbulent flow at Re τ = 100 and Re τ = 180. Wall transpiration (unsteady blowing/suction) with zero net mass flux is used as the control. The considered scheme results in the relaminarization of the turbulent flow at Re τ = 100 and a reduction in the drag force of about 50% at Re τ = 180. The optimal control results obtained using LES as approximate model of the flow indicate that LES can be efficiently used as a reduced order model of the Navier Stokes equations, to develop reliable control strategies at high Reynolds numbers and for more complex flow configurations.
Optimal feedback control of turbulent channel ow
1993 Annual Research Briefs, …, 1993
Feedback control equations have been developed and tested for computing wallnormal control velocities to control turbulent ow in a channel with the objective of reducing drag. The technique used is the minimization of a\ cost functional" which is constructed to represent ...
Application of reduced-order controller to turbulent flows for drag reduction
Physics of Fluids, 2001
A reduced-order linear feedback controller is designed and applied to turbulent channel flow for drag reduction. From the linearized two-dimensional Navier–Stokes equations a distributed feedback controller, which produces blowing/suction at the wall based on the measured turbulent streamwise wall-shear stress, is derived using model reduction techniques and linearquadratic-Gaussian/loop-transfer-recovery control synthesis. The quadratic cost criterion used for synthesis is composed of the streamwise wall-shear stress, which includes the control effort of blowing/suction. This distributed two-dimensional controller developed from a linear system theory is shown to reduce the skin friction by 10% in direct numerical simulations of a low-Reynolds number turbulent nonlinear channel flow. Spanwise shear-stress variation, not captured by the distributed two-dimensional controller, is suppressed by augmentation of a simple spanwise ad hoc control scheme. This augmented three-dimensional c...
Laminar and turbulent comparisons for channel flow and flow control
Journal of Fluid Mechanics, 2007
A formula is derived that shows exactly how much the discrepancy between the volume flux in laminar and in turbulent flow at the same pressure gradient increases as the pressure gradient is increased. We compare laminar and turbulent flows in channels with and without flow control. For the related problem of a fixed bulk-Reynolds-number flow, we seek the theoretical lowest bound for skin-friction drag for control schemes that use surface blowing and suction with zero-net volume-flux addition. For one such case, using a crossflow approach, we show that sustained drag below that of the laminar-Poiseuille-flow case is not possible. For more general control strategies we derive a criterion for achieving sublaminar drag and use this to consider the implications for control strategy design and the limitations at high Reynolds numbers.
2014
In this article, a forced reduced-order modeling approach, suitable for active optimal control of uid dynamical systems, based on the Proper Orthogonal Decomposition and perturbation method on the Reynolds-Averaged Navier-Stokes equations, is presented. Numerical simulation of turbulent ow equations is too costly for the purpose of optimization and control of unsteady ows. As a result, the POD/Galerkin projection and perturbation method on the RANS equations is considered. Using the perturbation method, the controlling parameter shows up explicitly in the forced reduced-order system. The feedback control of the controlling parameter is one of the objectives of this study. With the perturbation method, the e ect of the controller is sensed by the uid ow at each time step. The e ectiveness of this method has been shown on optimal control of the re-circulation problem for a turbulent ow over a step with blowing/suction controlling jets. Actuators are positioned at two di erent locations; blowing/suction jets at the foot and edge of the step, and blowing/suction jets at the wall of the step. Results show that the perturbation method is fast and accurate in estimating the re-circulated turbulent ow over a step. It is concluded that blowing/suction jets at the wall of the step are more e cient in mitigating ow separation.
Reynolds number effect on drag control via spanwise wall oscillation in turbulent channel flows
Physics of Fluids, 2019
The effect of Reynolds number (Reτ) on drag reduction using spanwise wall oscillation is studied through direct numerical simulation of incompressible turbulent channel flows with Reτ ranging from 200 to 2000. For the nondimensional oscillation period T + = 100 with maximum velocity amplitude A + = 12, the drag reduction (DR) decreases from 35.3% ± 0.5% at Reτ = 200 to 22.3% ± 0.7% at Reτ = 2000. The oscillation frequency ω + for maximum DR slightly increases with Reτ, i.e., from ω + ≈ 0.06 at Reτ = 200 to 0.08 at Reτ = 2000, with DRmax = 23.2% ± 0.6%. These results show that DR progressively decreases with increasing Reτ. Turbulent statistics and coherent structures are examined to explain the degradation of drag control effectiveness at high Reτ. Fukagata, Iwamoto, and Kasagi analysis in combination with the spanwise wavenumber spectrum of Reynolds stresses reveals that the decreased drag reduction at higher Reτ is due to the weakened effectiveness in suppressing the near-wall large-scale turbulence, whose contribution continuously increases due to the enhanced modulation and penetration effect of the large-scale and very large-scale motions in the log and outer regions. Both the power-law model (DR ∝ Re −γ τ) and the log-law model [DR = f (Reτ, ΔB), where ΔB is the vertical shift of the log-law intercept under control] are examined here by comparing them with our simulation data, from these two models we predict more than 10% drag reduction at very high Reynolds numbers, say, Reτ = 10 5 .
Linear optimal control of transient growth in turbulent channel flows
Acta Mechanica Sinica, 2019
This work investigates the suppression of linear transient growth in turbulent channel flows via linear optimal control. Two control algorithms are employed, i.e. the linear quadratic regulator (LQR) control based on full information of flow fields, and the linear quadratic Gaussian (LQG) control based on the information measured at walls. The influence of these controls on the development of both small-scale and large-scale perturbations is considered. The results show that the energy amplification of large-scale perturbations is significantly suppressed by both LQR and LQG controls, while small-scale perturbations are affected only by LQR control. The effects of the weighting parameters and control price on control performance are also analyzed for both controls, which reveals that different weighting parameters in the cost function do not qualitatively change the evaluation of control performance. As the control price increases, the effectiveness of both controls decreases markedly. For small-scale perturbations, the upper limit of the effective range of control price is lower than that for large-scale perturbations. When the Reynolds number is increased, it indicates that both LQR and LQG control become more effective in suppressing the energy amplification of large-scale perturbations.
Drag reduction via opposition control in a compressible turbulent channel
Physical Review Fluids, 2021
The compressibility effect on opposition drag control is studied via direct numerical simulation of turbulent channel flows at a bulk Reynolds number Re b = 3000 for three different bulk Mach numbers: M b = 0.3, 0.8, and 1.5. For all M b , the drag reduction (DR) has a similar trend as that of the strictly incompressible case; namely, DR first increases and then decreases with the sensing plane location y + d. With increasing M b , DR slightly decreases at small y + d but increases at large y + d. Consequently, y + d for achieving maximum drag reduction (DR max) shifts to larger values, namely, from y + d = 12.5 for M b = 0.3 to 20 for M b = 1.5, consistent with the outward shift of the peaks of Reynolds stresses at higher M b. By rescaling the sensing plane with semilocal units, a better collapse of DR is achieved among different M b , particularly for small y * d. The optimal sensing plane is found to be y * d ≈ 15 with DR max ≈ 23%. Interestingly, for large y + d cases, a resonance buffer layer characterized by a streamwise periodic array of spanwise-coherent rollers is established, one of the main reasons for the deterioration of drag reduction performance. This layer of hydroacoustic instability resonance results from the intense interaction of wall-normal wave propagation with the background mean shear. Space-time correlation of wall-normal velocity reveals near-wall organized spanwise structures with a well-defined streamwise wavelength λ x , decreasing with increasing M b .
Active turbulence control for drag reduction in wall-bounded flows
Journal of Fluid Mechanics, 1994
The objective of this study is to explore concepts for active control of turbulent boundary layers leading to skin-friction reduction using the direct numerical simulation technique. Significant drag reduction is achieved when the surface boundary condition is modified to suppress the dynamically significant coherent structures present in the wall region. The drag reduction is accompanied by significant reduction in the intensity of the wall-layer structures and reductions in the magnitude of Reynolds shear stress throughout the flow. The apparent outward shift of turbulence statistics in the controlled flows indicates a displaced virtual origin of the boundary layer and a thickened sublayer. Time sequences of the flow fields show that there are essentially two drag-reduction mechanisms. Firstly, within a short time after the control is applied, drag is reduced mainly by deterring the sweep motion without modifying the primary streamwise vortices above the wall. Consequently, the high-shear-rate regions on the wall are moved to the interior of the channel by the control schemes. Secondly, the active control changes the evolution of the wall vorticity layer by stabilizing and preventing lifting of the spanwise vorticity near the wall, which may suppress a source of new streamwise vortices above the wall.
Streamwise oscillation of spanwise velocity at the wall of a channel for turbulent drag reduction
Physics of Fluids, 2009
Steady forcing at the wall of a channel flow is studied via DNS to assess its ability of yielding reductions of turbulent friction drag. The wall forcing consists of a stationary distribution of spanwise velocity that alternates in the streamwise direction. The idea behind the forcing builds upon the existing technique of the spanwise wall oscillation, and exploits the convective nature of the flow to achieve an unsteady interaction with turbulence.