High-Resolution Large-Eddy Simulation of the Hurricane Boundary Layer

Understanding the dynamics of hurricane boundary layers is important for many reasons. One example is that being able to well characterize the turbulent wind field can help engineers better design offshore and near-shore structures that are subject to hurricane winds. Another example is that the turbulent transport processes that occur within the hurricane boundary layer (including air-sea interaction) are integral to the dynamics of the entire storm. With an improved understanding of these turbulent transport processes, atmospheric scientists will be better equipped to improve planetary boundary layer and surface layer schemes used in numerical weather prediction tools used to forecast hurricanes, ultimately leading to improved forecasting.

Although taking measurements within the hurricane boundary layer is extremely important in understanding hurricane boundary layer dynamics, it is an especially difficult task. A powerful tool to complement measurements is large-eddy simulation. Large-eddy simulation is a turbulence-resolving method for simulating fluid flow. It was born out of the atmospheric community in the 1970s and has proven an invaluable tool to study atmospheric boundary layer dynamics and processes. Only recently has it been applied to the hurricane boundary layer, and the results show great promise.

Our work is motivated by the planned installation of many wind farms along the hurricane-prone U.S. Atlantic coast. There is a great need to well characterize the turbulent wind-field of hurricane boundary layers to assess the mechanical loads that wind turbines will experience during the passage of a hurricane. Our ultimate goal is to be able to use large-eddy simulation to create such wind fields in any part of an idealized or real hindcast hurricane. We desire that our tools used to perform this task are validated against existing field data, including high-resolution radar data that has the ability to resolve turbulent structures.

Our atmospheric/renewable-energy large-eddy simulation code is unique in that it can employ adaptive mesh refinement, it has the ability to simulate wind turbine aero-servo-elastic behavior, it is based on the finite-volume formulation so it need not employ laterally periodic domains required in pseudo-spectral solvers, and it is performance portable and able to run on traditional CPU-based or emerging GPU-based high-performance computing systems.

We apply to this hurricane research our past research experience in offline mesoscale-microscale coupling between a mesoscale tool and our atmospheric/renewable-energy large-eddy simulation tools for non-hurricane situations. We employ two distinct methods for performing large-eddy simulation of the hurricane boundary layer, which we term “internal” and “boundary” coupling.

With internal coupling, we employ a laterally periodic domain that includes special source terms to account for large-scale radial gradients in the flow and centrifugal acceleration following the method outlined by Bryan et al. (2017). Internal coupling is powerful in its simplicity, but it is limited to regions significantly outside the hurricane eyewall.

A more general method able to capture even the complex eyewall region is boundary coupling. With boundary coupling, we treat our atmospheric/renewable-energy large-eddy simulation code as an offline nest within a WRF mesoscale simulation of a hurricane. The large-eddy simulation tool will be forced at the boundaries and through internal source terms using the WRF data.

The boundary-coupled method is the more challenging method. Key challenges include the need for inflow/outflow boundary conditions capable of having mixed regions of inflow and outflow that vary in time on a single domain boundary, and a means to rapidly “spin-up” resolved turbulence given inflow with no resolved turbulence from a mesoscale solver.

Our initial work has involved the implementation and testing of internal coupling using Bryan et al.’s (2017) method. Our results agree well with those of Bryan et al. An example overview image of the instantaneous flow simulated with this method is shown in Figure 1, which shows two different vertical slices of the instantaneous tangential flow field within a 5 km x 5 km x 2 km simulation domain. In this case, the mean wind speed 100 m above the surface is roughly 32 m/s. This domain is located 40 km from the center of the storm. Our near-term future work is focused on implementing and testing the boundary coupled method.

Figure 1: Two different vertical slices of the instantaneous tangential flow field within a 5 km x 5 km x 2 km large-eddy simulation of the hurricane boundary layer using Bryan et al.’s (2017) internal coupling method implemented in our atmospheric/wind-energy large-eddy simulation code.

References:

Bryan, G. H., Worsnop, R. P., Lundquist, J. K., and Zhang, J. A., “A Simple Method for SimulatingWind Profiles in the Boundary Layer of Tropical Cyclones,” Boundary-Layer Meteorology, Vol. 162, pp. 475–502, 2017.