TIME: 3:30 PM

LOCATION: GMCS 314

SPEAKER: Qi Wang, San Diego State University, Aerospace Engineering

ABSTRACT: Turbulence governs fluid motion across a wide range of spatial and temporal scales, posing fundamental challenges for modeling, estimation, and control. Accurate reconstruction of turbulent flow states from limited observations is essential for both scientific understanding and engineering applications. While laboratory experiments and numerical simulations each provide partial insight, their optimal integration remains a central challenge. We present a physics-constrained framework that systematically incorporates experimental measurements into computational flow models through adjoint-based optimization. The resulting inverse problem is formulated as a constrained optimization problem, with gradients obtained from the discrete adjoint of the governing equations. This approach enhances model fidelity, reduces uncertainty, and enables accurate flow state estimation from sparse data. We demonstrate the method in a turbulent channel flow, reconstructing the flow field using limited measurements.

Beyond state estimation, we show that the adjoint sensitivity and the associated Hessian naturally define the first- and second-order sensitivities of measurements, which together characterize the domain of dependence of a sensor. This perspective provides a powerful interpretation of measurement data and reframes flow-event identification as a geometric search problem in space and time. We apply this formulation to several canonical problems, including localization of a steady scalar source, identification of vortex sources in potential and laminar flow around an airfoil, and detection of heat sources in laminar stratified channel flows. Finally, we examine how adjoint sensitivities in turbulent flows expose fundamental limitations in flow reconstruction from sparse measurements, particularly as the estimation time horizon increases. These results highlight intrinsic challenges in turbulence estimation and provide possible pathways for future data assimilation strategies by stabilizing the adjoint fields.

BIO: Dr. Qi Wang’s research lies at the intersection of turbulence, inverse problems, and physics-informed machine learning. He develops physics-constrained and adjoint-based methods for interpreting sparse measurements and reconstructing complex flow dynamics, with applications ranging from scalar and vortex source localization to turbulent flow reconstruction and hypersonic boundary-layer diagnostics.

HOST: Jose Castillo