Meteorological Model/Engineering LES Hybrid Approach: Towards the Risk Assessment of Urban Areas in Strong Wind Events

Introduction

Strong wind events relating to meso- and micro-disturbances in the atmospheric boundary layer sometimes cause unpredicted disasters in urban and residential areas. Correct estimation of maximum wind velocity and analysis of unsteady turbulent flow behavior are keys for risk mitigation in those severe wind events from viewpoint of wind engineering. In order to realize the risk mitigation, it is essential to accurately capture high frequency component of velocity as well as general aspect and time development of meteorological phenomena. Numerical wind risk assessment of the urban district has been usually performed in a way reproducing the conditions of wind tunnel experiments. This conventional method is, however, not simply applicable to cases considerably influenced by meteorological processes and such meteorological phenomena itself because the severe wind events are often highly unsteady and spatially small and it does not satisfy the assumption that meteorological-scale physics are negligible in time and space compare with urban-district-scale physics.

There are several previous studies on application of meteorological model to high-resolution large eddy simulation (LES) in near ground region. Lundquist et al. [1] presented a method for introducing real complex three-dimensional urban terrain to the WRF model and demonstrated it with real urban district geometry in grid resolution of 2 m using ideal neutral boundary layer. Mirocha et al. [2] implemented nonlinear sub-grid turbulence stress model for WRF which can represent backscattering in the energy cascade and consider the effect of fine fluctuations in the wind flow.

Despite useful achievements of previous works, in most case, numerical meteorological models suffer from energy density decaying in high wave number or high frequency component possibly because of a number of model filters as explained in Skamarock’s works [3]. Besides, the tendency is also true for WRF-LES simulation from our experience and this can be a setback for the multi-scale simulation using meteorological model. Also, “gray zone” in the atmospheric boundary layer discussed by Wyngaard [4] requires caution when we perform multi-scale simulation using the meso meteorological model.

Methodology

We aim to overcome this setback and realize multi-scale simulation through three-step meteorological model/engineering LES hybrid approach shown in Figure 1 (a).

The first step is meteorological model simulation where cloud physics, baroclinity, and many other meteorological factors are considered. In the present study, we employed convective boundary layer generated WRF-LES in 50 m resolution for numerical validation and Tsukuba Tornado data simulated with the meteorological model JMANHM by Mashiko [5] converted into 50 m resolution for strong wind event application. In the second step, we regenerate high frequency velocity components of the meteorological model using our original LES-based method, in grid resolution of 25 m for the present study, and obtain broad-banded turbulence [6]. Comparing to the method for generating high frequency component of turbulence proposed Kawai et al. [7], which is suitable for typical convective boundary layer, this method can be applied arbitral types of flow field, because it does not require any precursor calculation and high frequency components are generated immediately inside the three-dimensional calculation domain. In the process of the turbulent energy cascade to finer scale, the present numerical simulation in this step deals with LES formulation using the sub-grid scale modeling, but does not employ the boundary-layer type of modeling of meteorological model. Thus, our case does not encounter the “gray zone” problem.

In the last step, using large eddy simulation (LES), we calculate unsteady flow structures in the near ground region with urban geometry model in the strong wind event relating to the meteorological disturbance. The data from previous step are used for initial condition and conditions for several boundaries in calculation process. We also estimate the pressure acting on the building and other data useful to wind risk assessment.

Summary of Results

For numerical validation, we confirmed that fine turbulent structures are generated throughout the domain during the calculation process using flow field data of meteorological model in convective boundary layer. Figure 1 (b) revealed the turbulence spectrum band of stream-wise wind velocity broadened with good fit to -5/3 Kolmogorov law in the inertial subrange.

We applied the present method to an extremely strong wind event of tornado. Figure 1 (c) is vertical wind speed at horizontal plane at Z = 250 m and it shows that method also successfully worked for actual meteorological case. As shown in Figure (d), clear velocity and pressure structure near-ground region around the complex urban geometries was resolved by LES with the real meteorological data, although we calculate only velocity in this domain and meteorological wind profile data below ca. 76 m for the initial condition and boundary conditions were interpolated assuming a power-law type profile. Figure 1 (e) reveals low pressure of tornado above the urban geometries broke into many parts and interacted significantly with vortices arise from corners of the buildings.

Reference:

[1] Lundquist, K., et al., An Immersed Boundary Method Enabling Large-Eddy Simulations of Flow over Complex Terrain in the WRF Model, Mon. Wea. Rev., vol 140, pp. 3936-3955, 2012

[2] Mirocha, J. D., et al., Implementation of a Nonlinear Subfilter Turbulence Stress Model for

Large-Eddy Simulation in the Advanced Research WRF Model, Mon. Wea. Rev., vol 138, pp. 4212-4228, 2010

[3] Skamarock, W. C., Evaluating Mesoscale NWP Models Using Kinetic Energy Spectra, Mon. Wea. Rev., vol.132, pp. 3019-3032, 2004

[4] Wyngaard, J. C., Toward Numerical Modeling in the ‘‘Terra Incognita’’, J. Atmos. Sci., vol. 61, pp 1816-1826, 2004

[5] Mashiko, W., A Numerical Study of the 6 May 2012 Tsukuba City Supercell Tornado. Part II: Mechanisms of Tornadogenesis, Mon. Wea. Rev., vol.144, pp. 3077-3098 (2016)

[6] Kawaguchi, M., et al., Generation of Higher Frequency Components for Wind Gust by Fusion Analyses of WRF and LES —Cases of CBL, Tornado and Urban Canopy Flow, Proceeding of APCWE9, 2017

[7] Kawai, H., High Frequency Recovering Technique of Turbulent Inflow for LES of Urban Wind, Proceedings of ICUC9, 2015