Abstract:We present a numerical study of two utility-scale 5-MW turbines separated by seven rotor diameters. The effects of the atmospheric boundary layer flow on the turbine performance were assessed using large-eddy simulations. We found that the surface roughness and the atmospheric stability states had a profound effect on the wake diffusion and the Reynolds stresses. In the upstream turbine case, high surface roughness increased the wind shear, accelerating the decay of the wake deficit and increasing the Reynolds stresses. Similarly, atmospheric instabilities significantly expedited the wake decay and the Reynolds stress increase due to updrafts of the thermal plumes. The turbulence from the upstream boundary layer flow combined with the turbine wake yielded higher Reynolds stresses for the downwind turbine, especially in the streamwise component. For the downstream turbine, diffusion of the wake deficits and the sharp peaks in the Reynolds stresses showed faster decay than the upwind case due to higher levels of turbulence. This provides a physical explanation for how turbine arrays or wind farms can operate more efficiently under unstable atmospheric conditions, as it is based on measurements collected in the field.

In this paper, we present simulation results for the three-dimensional, shock-driven Kelvin-Helmholtz instability. Simulations are performed with a Mach 2.0 shock propagating through a finite-diameter cylindrical column of dense gas inclined at an angle θ with respect to the shock plane. After passage of the shock, the gas curtain has accelerated along its axis and a Kelvin-Helmholtz instability forms on the column surface. This is the first known numerical reproduction of this phenomena in three dimensions, which has previously been observed in experiments with an inclined cylindrical gas column. The effects of changes to initial column angle (θ = 0 • , 10 • , 20 • , 30 • ) are explored in detail to complement experimental data. The effects of shock reflection near the base of the column are also examined to identify a possible flow perturbation near the foot which was seen in our previous two-dimensional numerical studies of shock-accelerated inclined gas curtains. The overall flow morphology compares well with experimental data in a cross-sectional plane through the column midpoint and a vertical plane through the column axis. Simulations were performed with FIESTA, an exascale ready, GPU accelerated compressible flow solver developed at the University of New Mexico.

Executive summaryThe scope of the work as proposed to the funding agency was to carry out numerical calculations of high-speed mixing flows using GASP, a computational fluid dynamics (CFD) code. The ultimate goal is to use GASP for prediction and optimization of chemical laser flows. To do it, one must have confidence in the performance of the hydrodynamic and mixing part of the code.For complex numerical simulations, it is essential to perform code validation -a quantitative comparison of numerical results with experimental results confirming that the code faithfully reproduces the physics of the real problem. Two detailed validation exercises were performed with GASP:" Simulation of a shock-accelerated mixing flow. This validation exercise compared the numerical results produced by GASP with highly-resolved experimental data on flows subject to Richtmyer-Meshkov instability (RMI), the preferred test problem for quantitative assessment of the performance of CFD codes in prediction of the properties of high-speed mixing flows. The validation problem revealed that GASP can faithfully reproduce all the large-scale quantitative properties of the flow. We also gained important additional insights into the strengths and limitations of GASP (and other numerical codes) in prediction of disordered small scales in flows transitioning to turbulence." General jet-in-crossflow problem. The problem of a supersonic jet discharging into a supersonic crossflow presents a significant computational challenge, yet provides a valuable validation exercise when the results are compared with experimental data. The conditions for this exercise were chosen to achieve two goals: -Provide comparison with best experimental data available in open literature.-Test the code in a generic configuration that can be easily modified to represent a specific chemical laser nozzle injection scheme.The agreement with the experimental data was quite good, yet subtle features of the flow that were not apparent in the experiment were revealed by the simulation.The results of our work are as follows."* We found the hydrodynamic model used in GASP to be suitable for prediction of mixing in chemical laser flows."* We found the error in prediction of the geometry of large-scale features (e.g., counter-rotating vortex pairs) not to exceed 10-15% when compared with experiment.