Direct Numerical Simulation of Compressible, Turbulent Flow

PI Name: 
Jonathan Poggie
PI Email: 
Jonathan.Poggie@gmail.com
Institution: 
Purdue University
Allocation Program: 
INCITE
Allocation Hours at ALCF: 
150 Million
Year: 
2016
Research Domain: 
Engineering

During long-duration, supersonic or hypersonic flight in the atmosphere, the vehicle must withstand both intense heating and unsteady mechanical loads. A fundamental difficulty in this regime is the presence of long time-scale (1–100 ms), low-frequency (10–1000 Hz) pressure fluctuations under separated, turbulent boundary layers. These fluctuations lie in a regime near the typical resonant frequency of aircraft panels, and thus lead to severe structural fatigue loading. A key scientific question remains as to why such low-frequency oscillations exist.

The disparity of length and time scales between fine-grain turbulence and large-scale flow unsteadiness makes computational simulation of the thermal and mechanical loads on high-speed aircraft inherently challenging. Focusing on the basic science of unsteady separation in compressible, turbulence flow, the aim of this project is to investigate perturbed, supersonic turbulent boundary layers through massively-parallel, direct numerical simulations.

Using the high-order, finite-difference code HOPS (Higher Order Plasma Solver), the research team is employing a compression ramp configuration to generate flow separation—a configuration representative of aircraft structures. The main objective is to test the validity of the amplifier and oscillator models of separation unsteadiness by comparing the wall pressure spectra near separation for a turbulent incoming boundary layer and a laminar incoming boundary layer under the same flow conditions.

In this multi-year project, simulations will be carried out in four stages, with the first year focused on preliminary studies in preparation for production work. Coarse-grid simulations will determine grid resolution requirements and suitable flow conditions, such as ramp angle and Reynolds number. Eventually, this work will lead to fine-grid simulations of incoming turbulent flow meant to replicate published experimental data and explore the possibility of mitigating unsteadiness with flow control.