Speaker
Description
High-lift airfoils operating at low Reynolds numbers are governed by laminar boundary-layer behavior, in which laminar separation bubbles significantly influence lift generation and overall aerodynamic performance. In the existing literature, laminar separation bubbles are generally considered an inherent feature of low-Reynolds-number operation and are often associated with reduced performance and increased sensitivity to operating conditions [2,3]. As a result, classical high-lift airfoil design strategies primarily focus on delaying, mitigating, or minimizing flow separation rather than exploiting it. In the present work, a different design paradigm is explored. A single-element high-lift airfoil is optimized directly at a chord-based Reynolds number of 100,000 using a single-step Deep Reinforcement Learning framework for direct shape optimization, following the approach proposed in [1]. XFOIL is employed as the aerodynamic solver within the optimization loop, enabling efficient evaluation of aerodynamic performance and boundary-layer behavior. Within this framework, the laminar separation bubble is not suppressed but intentionally exploited as a design feature. The optimization promotes and shapes the separation bubble to maximize upper-surface suction levels, leading to lift performance exceeding that of classical low-Reynolds-number high-lift airfoils [2,3].
The optimization is formulated as a multipoint problem around an angle of attack of 4 degrees, with additional conditions at 3 and 5 degrees. Aerodynamic performance is optimized directly, without imposing boundary-layer constraints, resulting in a geometry that sustains a persistent laminar separation bubble in the forward chord region.
The optimized design is investigated experimentally using a combined approach based on planar Particle Image Velocimetry and quantitative Infrared Thermography. Particle Image Velocimetry provides characterization of the external flow field, enabling identification of separation, transition, and reattachment, as well as the associated turbulent kinetic energy distribution. In parallel, quantitative Infrared Thermography is used to reconstruct convective heat transfer through a thin-film sensor model, allowing the derivation of Stanton number distributions.
Separation and reattachment locations inferred from the Stanton number fields are compared with Particle Image Velocimetry measurements and show strong mutual consistency. In particular, reattachment regions identified by peaks in the Stanton number match local maxima of turbulent kinetic energy. Through the Reynolds analogy, the heat-transfer measurements provide indirect access to wall shear stress behavior [4], confirming the physical consistency between experimental observations and XFOIL predictions. The results demonstrate that laminar separation bubbles could be deliberately designed and controlled through airfoil geometry to achieve robust high-lift performance at low Reynolds numbers.
References
[1] Viquerat, J., Rabault, J., Kuhnle, A., Ghraieb, H., Larcher, A., & Hachem, E. (2021). Direct shape optimization through deep reinforcement learning. Journal of Computational Physics, 428, 110080.
[2] Selig, M. S., Low Reynolds Number Airfoil Design, von Kármán Institute for Fluid Dynamics LectureSeries,2003.
[3] Liebeck, R. H. (1978). Design of subsonic airfoils for high lift. Journal of aircraft, 15(9), 547-561.
[4] Carlomagno, G. M., & Cardone, G. (2010). Infrared thermography for convective heat transfer measurements. Experiments in fluids, 49(6), 1187-1218.
