SOFT

Simulation, Optimisation, Fabrication and Testing

Marc Garbey, Malek Smaoui, Nicolas De Brye and Christophe Picard


Results

Simulation:
The first set of simulations allowed an exploration of the physical problem and its numerical resolution:
  • We obtain the rotation speed of the VAWT when performing unsteady simulations -- highly time consuming.
  • We calculate the average torque exerted on the blade for different angular positions using only steady simulation -- less execution time.
Comparing the above simulations, we show that we may obtain the same optimization results when maximizing "the mean torque in steady" instead of "the rotation speed in unsteady".
We save on execution time.

Optimization:
  • Coupled with ADINA, the genetic algorithm can only run sequentially.
  • Running the genetic algorithm with small generation size and small number of generations is not conclusive and increasing these parameters is not viable due to the excessive execution time.

Future and ongoing work

  • Implementing a fast and portable FSIS tool.
  • Parallelizing the genetic algorithm to run on BOINC (Berkeley Open Infrastructure for Network Computing) using the new FSIS tool for the evaluation step.
  • Defining more complex parametrical geometric models of the windmill shape to be used in the next iterations of the SOFT loop.

<< Prev  1  2  3  4  5  6