Traveling Wall Waves Cut Turbulent Drag by Up to 66.6% in Supersonic Channel Flow

Engineers have long chased ways to shave friction drag off the surfaces of aircraft and other vehicles moving through air, because skin friction accounts for more than half of the total drag on commercial aircraft. A new computational study now maps, in unusual detail, how one active control technique behaves when the flow is no longer safely incompressible. Using direct numerical simulations, researchers examined streamwise-traveling waves of spanwise velocity imposed at the walls of a turbulent channel flow at three bulk Mach numbers, 0.3, 0.8 and 1.5, all at a fixed bulk Reynolds number of 3000. The result is a frequency-wavenumber map of drag variation that shows where the waves reduce friction drag, where they perversely increase it, and how compressibility shifts the entire picture.

The technique builds on one of the simplest active control strategies known: oscillating the wall in the spanwise direction. That method, characterized by just an amplitude and an oscillation period, was shown in early work to suppress turbulence most effectively at a nondimensional period of 100 wall units, achieving drag reduction of up to 40 percent. The traveling-wave extension adds a wavelength, letting the wall motion propagate downstream as a wave of spanwise velocity rather than oscillating in place everywhere simultaneously. The phase speed of that wave, the ratio of angular frequency to wavenumber, turns out to be the decisive parameter: depending on its value, the same control can either weaken or strengthen the near-wall turbulence regeneration cycle that generates most of the friction drag.

In the new simulations, performed with the GPU-based STREAms solver in a computational box of 6π by 2 by 2π channel half-heights, the wall velocity amplitude was fixed at 12 wall units, with nondimensional periods from 25 to 300 and wavenumbers from 0 to 0.015. Sixty-seven simulations were run for each Mach number under constant-flow-rate conditions, meaning the pressure gradient adjusted over time to keep the bulk velocity fixed. Drag reduction was quantified by comparing the time-averaged skin friction coefficient of each controlled case against the uncontrolled flow at the same Reynolds and Mach numbers. The solver was validated by reproducing earlier spanwise-oscillation results at Mach 0.3 and 1.5 with excellent agreement.

The headline finding is that the drag variation trends look strikingly similar across all three Mach numbers, but the magnitudes grow with compressibility. At Mach 0.3, the reference case for incompressible flow, the maximum drag reduction reached approximately 48.53 percent at a specific frequency-wavenumber combination, surpassing the 34.8 percent achieved by a simple spanwise wall oscillation and comparable to roughly 50 percent reductions reported previously in incompressible studies. At Mach 0.8 the maximum rose to about 52.59 percent. At Mach 1.5, the flow actually relaminarized under optimal forcing, yielding a maximum drag reduction of 66.6 percent, an outcome the authors attribute to the relatively low friction Reynolds number of about 200 placing the flow in a transitional regime.

The dark side of the map is equally systematic. At moderate phase speeds, the waves amplify drag instead of reducing it. Maximum drag increase climbed from 20.22 percent at Mach 0.3 to 23.12 percent at Mach 0.8 and 31.32 percent at Mach 1.5. The drag-increasing band occupies a narrow range of phase speeds, from about 8.29 to 13.22 wall units at Mach 0.3, expanding to 9.21 to 15.96 at Mach 1.5, so compressibility pushes this harmful region toward higher phase speeds. Across the full parameter space, drag follows a reduction-increase-reduction pattern as phase speed rises: strong reduction at low speeds, substantial increase at moderate speeds, and renewed reduction at high speeds.

Why does a moving wall sometimes backfire? The turbulence statistics tell the story. In drag-reduction cases, the peak of the streamwise Reynolds stress drops and shifts away from the wall, in the extreme case moving into the logarithmic layer, while the wall-normal stress behaves similarly. Quadrant analysis, which decomposes the Reynolds shear stress into four types of velocity fluctuation events, shows that ejections of slow fluid away from the wall and sweeps of fast fluid toward it, the two events that together supply roughly 70 percent of the shear stress in the uncontrolled flow at Mach 1.5, are strongly suppressed when drag falls. At the maximum drag-reduction case the fluctuating velocities were nearly zero, the signature of relaminarization. In drag-increase cases the opposite occurs: sweeps and ejections intensify, vortex structures multiply until they nearly fill the channel, and the shear stress rises in direct proportion to the drag penalty.

One of the most intriguing discoveries concerns phase-locking. When the wave’s phase speed matches the convective velocity scale of the near-wall flow, about 10 wall units, the wall forcing and the flow structures lock into step, and streamwise streaks are cyclically tilted by the moving wall. Earlier studies had observed this phenomenon only in drag-increase regimes for incompressible flow and only in drag-reduction regimes for compressible flow. The present simulations found phase-locking in both regimes, at every Mach number tested. Notably, phase-locking sharply reduces streamwise fluctuations and boosts wall-normal fluctuations, yet the Reynolds shear stress changes only modestly, and the near-wall structures it produces appear to be a direct response to the wall motion rather than a reliable indicator of whether drag will rise or fall.

The frequency of the wall forcing also leaves a visible fingerprint on the streaks. Low-frequency oscillations imprint a periodic wave-like pattern in the streamwise direction that significantly modulates the streak structures, with the streak wavenumber matching that of the wall wave. High-frequency oscillations produce no such modulation. The explanation lies in the Stokes boundary layer: its thickness shrinks as frequency rises, so fast oscillations cannot reach high enough into the flow to perturb streaks that have a finite wall-normal extent.

The authors are careful about scope. These are direct numerical simulations of a canonical channel flow at a friction Reynolds number near 200, a value chosen to align with prior incompressible and compressible studies but low enough that transitional behavior, including full relaminarization at Mach 1.5, plays a role in the largest reported reductions. Still, the systematic mapping of how compressibility enhances both the benefits and the penalties of traveling-wave control, and the identification of phase speed as the master variable, gives designers of future flow-control systems a quantitative guide.

Subject of Research: Drag control of compressible turbulent channel flow using streamwise-traveling waves of spanwise wall velocity

Article Title: The effect of streamwise-traveling waves on the drag variation of compressible channel flow

Article References: Wang, B., Wu, R., Li, J., Liu, Y., & Zhou, Q. (2026). The effect of streamwise-traveling waves on the drag variation of compressible channel flow. Advances in Aerodynamics, 8(1), Article 17. https://doi.org/10.1186/s42774-025-00223-6

Image Credits: AI Generated

DOI: 10.1186/s42774-025-00223-6

Keywords: drag reduction, turbulent channel flow, compressible flow, direct numerical simulation, streamwise traveling waves, spanwise wall oscillation, Mach number, friction drag, phase-locking, coherent structures, Reynolds stress, fluid dynamics