Measurement-infused simulations of hypersonic transition on cones with flares
Abstract
Hypersonic boundary-layer transition on cones is a complex, chaotic phenomenon that is very sensitive to environmental uncertainties. Free-stream disturbances rst interact with the leading-edge shockandtheoutcomedependsontheightconditionsandcharacteristicsoftheincomingwaves. Downstream, waves that effectively breach the boundary layer may amplify, interact with one another nonlinearly, and spur new waves and distortions. The dynamics is further complicated in presence of cone ares. A positive are leads to a compression shock that may introduce local unsteady separation and the radiation of perturbation energy along the shock; a negative are leads to an expansion. The impact of corner effects on instability waves will depend on the frequency/wavenumber and amplitude of the instabilities, be they in the early linear regime, near saturation, undergoing secondary instability, or have formed wave packets and turbulence spots ahead of the corner. The transition mechanism is therefore the outcome of all these antecedent processes that are each sensitive to uncertainties, and these complexities frustrate both computational and experimental efforts: Computations have the ability to non-intrusively probe the full ow eld at full spatio-temporal resolution, but are not efcient at exploring a large parameter space and often invoke idealizations that can compromise accuracy (e.g. shock capturing methods). Experiments, on the other hand, can span many ow congurations efciently and are immune to the idealizations of computations, but measurements in the hypersonic regime are notoriously difcult. In this effort, we will adopt our measurement-infused simulation techniques to accurately relate transition on cones with ares to the incoming free-stream disturbances up stream of the leading-edge shock. The approach uses available observations, e.g. surface PCB or ALTP sensor data and Schlieren images, as constraints and adopts variational techniques to determine the unknown free-stream disturbances whose Navier-Stokes evolution reproduce the observations. Since this problem may admit non-unique solutions, we further constrain our predictions based on prior knowledge and characterizations of the ow environment. The ultimate outcome is a spatially and temporally resolved description of the full ow eld, from the free stream well into the turbulent boundary layer, that optimally decodes the sensor data to reveal the preceding events encoded in those signals. The analysis will yield unique understanding of how the leading-edge shock and corner shocks and expansions not only affect the evolution of instability waves, but also lter information and thus obscure prediction of the upstream conditions from downstream information, which has bearing of our understanding of the physical problem and the practical relevance in the context of interpretation of in-ight measurements. Our measurement-infused simulations adopt adjoint-and ensemble-variational techniques, which are robust, accurate, build on decades of advancements in data assimilation, and also require high level of expertise. We will also develop new tools that exploit machine learning techniques, including Physics-Informed Neural Networks (PINNs) that aim to reproduce the experimental measurements and satisfy the full Navier-Stoke sequations. We will apply PINNS to interpret data from transitional, hypersonic boundary layers on cones with ares, and our predictions will be compared to the benchmark results using variational techniques.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 15, 2021
- Source ID
- N000142112148
Entities
People
- Tamer A. Zaki
Organizations
- Johns Hopkins University
- Office of Naval Research
- United States Navy