Optimization and Measurement-Infused Simulations Of Three-Dimensional Separated Flows Around Inclined Bodies of Revolution
Abstract
SOWAccurate prediction of high Reynolds-number flow over a spheroid at incidence is challenging dueto the complex near-wall turbulence dynamics, or in the boundary layers (BL). The challenge isexacerbated when these layers separate from the surfacea phenomenon sensitive to incidence,pressure distribution, state of the upstream BL and turbulence structures, surface conditions andenvironmental disturbances. This project examines the flow around a 6:1 prolate spheroid at incidence,the onset of closed and open separation, the dynamics of the separated layers and reattachment.The target Reynolds number is ReL 270106 based on axial length, and the presentcomputational effort is closely coordinated with parallel experiments.The study will initially evaluate the role of trip on the resulting BL characteristics which ultimatelyimpact separation. Sensitivity analyzes and optimization of trip will target the downstream formationof equilibrium turbulent BL without memory of the forcing. For the spheroid simulations,the target Re is inaccessible by direct numerical simulations (DNS) and turbulence modeling cancompromise the fidelity of computational predictions. To address model deficiencies, variationaltechniques will be adopted to optimize large-eddy simulations (LES) for 3D separation on thespheroid. Reference data will be obtained from (i) DNS at moderate Re and (ii) measurements offlow statistics at various stations on the spheroid at design Re. The optimized LES can provide thefull flow field around the spheroid and in the wake, but they are nse deficiencies since they are physical realizations of Navier-Stokes, but measurements are difficultto perform and will be limited to flow statistics at specific locations and spatio-temporal data atonly one location in the wake. In order to overcome both the computational and experimental limitations,measurement-infused LES will be performed where the simulations will be constrainedto satisfy the measurements using variational techniques. The formulation is an inverse problem:starting from observations, determine the model, inflow and boundary parameters that best reproducethe measurements. While computationally challenging, such capability is transformative:data-assimilated LES predictions will follow the exact trajectory in state space that is probed inthe experiments, and thus enhance the fidelity and realism of the simulations beyond conventionalapproaches. In addition, the simulations will provide high-fidelity data away from the measurementsites, thus expanding the value of the experiments. In this approach, a measurements is nota mere record of the flow state at the specific spatial location and time instance, but a record ofthe surrounding flow in space and time, and our simulations decode that record to provide a higherresolution description of the flow than the measurements alone. With the new interpretation ofmeasurements in mind, the placement of sensors must be optimized not only to measure phenomenaof interest, but also to maximize the accuracy of the measurement-infused simulations. Thesame variational framework will be adopted to identify the optimal placement of sensors in theexperiments, in order to best achieve accurate simulations.The onset of separation will be examined, with particular attention to its sensitivity to turbulencestructures in the upstream boundary layer. This sensitivity can be computed using similar variationaltechniques to those adopted for measurement-infused LES. The sensitivity of the wake structureto the separation region will be studied in detail, and the results will provide an unprecedentedview of the dyna
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 20, 2020
- Source ID
- N000142012715
Entities
People
- Tamer A. Zaki
Organizations
- Johns Hopkins University
- Office of Naval Research
- United States Navy