Improved Mesoscale Simulation Toolkit for Wind Resource Assessment over Steep Complex Terrain

 

Currently, extracting energy from the wind has become one of the main and most reliable renewable sources to generate clean electricity. In 2012, the wind industry grew by 19 per cent worldwide, achieving a record level of 44,711 MW of new installed wind power capacity. Today, there are over 150,000 wind turbines operating around the world in over 90 countries [1]. Wind resource assessment constitutes one of the main areas of research and development in the wind industry. [2]. It is regarded as the first step for achieving an appropriate wind farm planning, by gathering sufficient amount of information of the site desired for the installation of wind turbines.

The fast growing wind industry need greater accuracy in mapping the wind climate near the ground in order to allow wind farm developers to identify areas of high wind energy potential [3]. They require highly accurate atmospheric boundary layer simulations over diverse topographic configurations for wind farm planning at the lowest possible cost. However, the accuracy of the wind resource assessment methods is limited, mostly due to inefficient numerical schemes and heavy computational capacity needed to resolve fine-scale wind structures close to the topographic surface.

Hence, the research paper describes the methodology followed by the researchers of the Canadian Research Laboratory for Nordic Environment Aerodynamics of Wind Turbines (NEAT) from École de technologie supérieure (ÉTS), Montreal, Quebec, Canada and Environment Canada to upgraded the mesoscale compressible community model (MC2). The proposed solution is part of the latest improvements made to the Wind Energy Simulation Toolkit (WEST), open-source software developed by the Recherche en Prévision Numérique (RPN) group of Environment Canada, for wind resource assessment.

The Wind Energy Simulation Toolkit (WEST) is built to estimate the wind resource by taking into account atmospheric phenomena of a wide range of scales, both in time (from decadal to diurnal variations) and in space (from synoptic large-scale to meso-/micro-scale). WEST was employed to obtain the complete wind resource mapping and statistical analysis compiled in the Canadian Wind Atlas.

The Canadian mesoscale compressible community model (MC2) is a limited area non-hydrostatic model, [4], [5]. [6]. It solves the fully elastic Euler equations and was originally developed as for weather research and forecasting. Since almost a decade, it became the core dynamics solver of WEST, coupled with MS-Micro and WaSP to account for the whole range of atmospheric flows over complex terrain. MC2 continues to be a fundamental reference for other semi-implicit semi-Lagrangian methods and community models which are oftenly employed in wind resource assessment [2].

Recently, the MC2 model has been upgraded with a state-of-the-art large eddy simulation (LES) methodology which increases its capabilities to resolve highly complex atmospheric flows aided with advanced sounding initialization, dynamic downscaling, 3D turbulence diffusion, sub-grid scale modeling, interpolation procedures and vertical grid staggering. This upgraded version yields excellent results for thermally stratified atmospheric flows. Consequently, this mesoscale model can be considered an optimal computational fluid-dynamics (CFD) solver that accounts for very important meteorological aspects on long- and short-term predictions of wind energy potential.

A research made by [3] demonstrate that WEST shows relatively strong correlations between its simulated long-term mean wind speed and the measurements from ten wind energy monitoring stations. However, in the mountainous terrain, WEST tends to predict wind speeds which are about 40% too high. The model also produces some wind direction shifting over high impact terrain, with evident misalignment respect to the geostrophic forcing.

Hence, the main objective of this research is the enhancement of the MC2 model to ensure a robust numerical stability and accurate results by reducing the numerical noise for wind resource assessment over complex terrain with moderate and steep slopes. This is achieved by redefining the buoyancy forcing to cast an efficient dynamic initialization by implicitly treating the non-linear terms, obtained after the discretization of the fully elastic Euler equations solved by the model’s dynamic kernel with higher order numerical schemes such as the semi-implicit semi-Lagrangian scheme (SISL) introduced by [7].

Until recently, the basic three-time level SISL scheme of the MC2 model was considered to be unconditionally stable, for the flat terrain case at least [5], [8]. In spite of the sophisticated elliptic solver (GMRES) that was implemented by [4] in MC2 to include all the terrain-induced metric terms in the semi-implicit (SI) branch of the solution procedure, the model remained subjected to high-frequency noise, known to be related to topography itself and to various types of filtering as described by [9]. Part of this noise is generated by imbalances in the initial temperature and pressure fields thought to be controlled by the introduction of an off-centering parameter in the SISL averaging operators.

The proposed solution to this numerical problem is a new semi-implicit (NSI) scheme based on a redefinition of the buoyancy. To obtain its final form, an analogous modal analysis is applied to the resulting space-time averaged equations, resulting in a much simpler dispersion relation. It turns out that according to this new scheme all the non-linear terms are treated implicitly, including the remaining divergence term of the thermodynamic equation. Consequently, the NSI scheme yields very stable results and reduces the numerical noise several orders of magnitude for simulations over steep topographic gradients. A simple comparison of vertical velocity fields is presented in Figure 2, obtained with different terrain slopes for an isothermal initially resting atmosphere case, helps measure the performance of the proposed solution to keep numerical noise to the lowest possible levels.

In conclusion, the upgrading process yields a more stable mesoscale model capable of accurately modeling wind flow over high resolution steep terrain, taking into account key meteorological aspects for long- and short-term predictions. The major contribution to science and wind engineering is a new space-time discretization scheme for mesoscale modeling that reduces the numerical noise several orders of magnitude for simulations with different parametric combinations.

For a more comprehensive discussion about the enhanced numerical methods for mesoscale modelling of wind-flow over complex terrain, we will invite you to read the Research Paper as soon as it will be available :

Alex Flores-Maradiaga, Robert Benoit, Claude Girard (2013) : On the Enhancement of Numerical Stability and Noise Control for Mesoscale Modelling over Steep Complex Terrain.