TY - GEN
T1 - Computational tool for a mini-windmill study with SOFT
AU - Garbey, M.
AU - Smaoui, M.
AU - De Brye, N.
AU - Picard, C.
PY - 2009
Y1 - 2009
N2 - In this paper, we present a parallel computational framework for the completely automated design of a Vertical Axis Fluid Turbine (VAFT). Simulation, Optimum design, Fabrication and Testing (SOFT) of the VAFT is integrated into a hardware/software environment that can fit into a small office space. The components of the four steps design loop are as follows Simulation: We use a parallel CFD algorithm to run a direct simulation of the fluid structure interaction problem. We derive from that computation the torque and the average rotation speed for a given friction coefficient on the rotor shaft and an average flow speed. Our objective is to get the most power out of the windmill, consequently the highest rotation speed possible. Optimization: We optimize the shape of the blade section with a genetic algorithm and/or a surface response. The evaluation of the objective function (average rotation speed) corresponds to the direct simulation of the Navier Stokes flow interacting with the rotating turbine, until reaching a stationary regime. Because this simulation is compute-intensive, we distribute the evaluation of the objective function for the different shapes (gene or parameter combinations) on a network of computers using an embarrassingly parallel algorithm. Fabrication: The optimization procedure results in a supposedly optimum shape in the chosen design space. This shape is sent to a 3-D printer that fabricates the real turbine. This turbine is set up such that it can be easily mounted on a standard base equipped with an electric alternator/generator. Testing: The windmill is tested in a mini wind tunnel. The electric output is measured and a video camera can directly monitor the windmill rotation through the transparent wall of the wind tunnel. This information can be analyzed by the computer system and comparison with the simulation is assessed. Figure 1 gives a graphical overview of the SOFT concept.
AB - In this paper, we present a parallel computational framework for the completely automated design of a Vertical Axis Fluid Turbine (VAFT). Simulation, Optimum design, Fabrication and Testing (SOFT) of the VAFT is integrated into a hardware/software environment that can fit into a small office space. The components of the four steps design loop are as follows Simulation: We use a parallel CFD algorithm to run a direct simulation of the fluid structure interaction problem. We derive from that computation the torque and the average rotation speed for a given friction coefficient on the rotor shaft and an average flow speed. Our objective is to get the most power out of the windmill, consequently the highest rotation speed possible. Optimization: We optimize the shape of the blade section with a genetic algorithm and/or a surface response. The evaluation of the objective function (average rotation speed) corresponds to the direct simulation of the Navier Stokes flow interacting with the rotating turbine, until reaching a stationary regime. Because this simulation is compute-intensive, we distribute the evaluation of the objective function for the different shapes (gene or parameter combinations) on a network of computers using an embarrassingly parallel algorithm. Fabrication: The optimization procedure results in a supposedly optimum shape in the chosen design space. This shape is sent to a 3-D printer that fabricates the real turbine. This turbine is set up such that it can be easily mounted on a standard base equipped with an electric alternator/generator. Testing: The windmill is tested in a mini wind tunnel. The electric output is measured and a video camera can directly monitor the windmill rotation through the transparent wall of the wind tunnel. This information can be analyzed by the computer system and comparison with the simulation is assessed. Figure 1 gives a graphical overview of the SOFT concept.
UR - http://www.scopus.com/inward/record.url?scp=78651560090&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-02677-5_32
DO - 10.1007/978-3-642-02677-5_32
M3 - Conference contribution
AN - SCOPUS:78651560090
SN - 9783642026768
T3 - Lecture Notes in Computational Science and Engineering
SP - 291
EP - 297
BT - Domain Decomposition Methods in Science and Engineering XVIII
T2 - 18th International Conference of Domain Decomposition Methods
Y2 - 12 January 2008 through 17 January 2008
ER -