How much electricity can a wind turbine produce? This depends greatly on its location. Finding an answer requires studying the seasonal prevailing wind patterns, air turbulence, and other variables for a specific geographic site under consideration. Weather data sets can be huge, leading to many challenges regarding modeling, simulation, and visualization of the problem.
Beneficiaries demand to know how much energy they will produce, and what their return on investment will be before they are installed. This is often time consuming, especially on complex terrain. Using EUMEDGRID e-Infrastructure and in-house, tailored, numerical models, a team of researchers, led by myself (Ashraf Hussein) at Ain Shams University, Egypt, can now answer these questions quickly enough to identify potential markets for wind energy, especially in Egypt, to meet aggressive renewable energy goals. The work is due to be published in the International Journal of Computational Methods in June.
A computational framework for modeling and simulation of micro-scale (and early meso-scale) wind farms using massively parallel high performance computing platforms has been conceived. The wind turbines installed in the wind farm are modeled by the actuator disk method using the Virtual Blade Model (VBM). This model considers the presence of the wind turbines' rotors implicitly through source terms in the momentum equations. In addition, an efficient parallel algorithm for implementing the model through the actuator disk method was developed and integrated with the parallel Computational Fluid Dynamics core simulation engine.
Forecasting the success of these wind farms can therefore be performed within the decision making time. In addition, important topics which are yet to be fully understood can also be studied and analyzed, such as the interaction between the wind turbines themselves and the terrain of the wind farm, independent of its complexity. The scalability analysis for the present framework was carried out using IBM Blue Gene/L with up to 1054 computing nodes. Following this, some real simulations were carried out using the EUMEDGRID e-Infrastructure through the grid interface provided by the Egyptian Universities Network (EUN), the Egyptian partner in EUMEDGRID-Support project.
This research proved that EUMEDGRID e-Infrastructure is currently one of the most important resources that researchers in Mediterranean countries can rely on to carry out such environmental simulations with large domains and/or sophisticated models.