An Analysis of Inflow Perturbation Strategies for Mesoscale-Driven Microscale Large-Eddy Simulations

There is a growing need for the ability to perform mesoscale-driven microscale atmospheric large-eddy simulations (LES). Historically, microscale atmospheric LES has been performed for canonical conditions in which periodic boundary conditions are used. The turbulent boundary layer forms given a geostrophic wind and surface forcing. This atmospheric LES strategy has proven valuable and can even be coupled with the mesoscale through time-varying geostrophic winds and surface fluxes, along with the inclusion of a large-scale advective term in the temperature evolution equation. However, the method is constrained to horizontally periodic conditions. For uniform conditions, this translates to horizontal homogeneity. Most generally, horizontally periodic or homogeneous conditions are not the case. This is true, for example, in regions of complex terrain, where there is land-sea interaction, or near a weather front. The driving application of this work is wind energy. Wind plants are often located in regions of complex terrain or their geographical extent is so large that mesoscale-driven heterogeneity exists.

A number of researchers, including the authors, perform wind-plant aerodynamics simulations by embedding a wind-turbine aerodynamics model (actuator disks or actuator lines) within a microscale atmospheric LES. There is growing need to be able to perform these wind-plant simulations under non-periodic, horizontally heterogeneous, mesoscale-forced situations. For example, there is a need to better understand how newly proposed wind-plant control systems behave during frontal passages. In performing forensic analysis of wind turbine failures, it is important to capture the mesoscale effect on the microscale winds. Mesoscale input will become essential to future wind plant power forecasting that may rely on microscale wind-plant simulations.

A method for performing non-periodic, mesoscale-driven wind-plant microscale LES is to use time and space-varying mesoscale information as the inflow boundary conditions to the wind plant LES. Because the smaller-scale turbulence in the mesoscale simulation is modeled and not resolved, these inflow boundary data do not contain resolved turbulence. On the other hand, the microscale simulation is turbulence resolving. The inflow data, therefore, must be perturbed in some way so that resolved turbulence forms as quickly as possible within the microscale domain and comes to equilibrium within a minimum distance. Without perturbation, this process can take tens of kilometers.

In this work, we compare various inflow perturbation strategies. Specifically, we will compare the performance of methods that perturb the inflow temperature field, the inflow velocity field, or that superimpose synthetic turbulence onto the mesoscale inflow. We will assess how quickly each of these methods form equilibrium turbulence under a few different atmospheric stability conditions.