Modelling energy output to optimize tidal turbine placement
LE3 .A278 2010
2010
Chipman, Hugh Ranjan, Pritam
Acadia University
Bachelor of Science
Honours
Mathematics and Statistics
Mathematics & Statistics
As home to the highest tides in the world, the Bay of Fundy, especially the Minas Channel, shows great potential for harnessing tidal power. An important step is to find the best locations for turbine fences that will maximize the amount of extractable kinetic power. Various mathematical models can be used in prediciting energy generated by tidal turbines. Thus they enable the power maximization to be carried out before placing real turbines. We choose the Power Vs Location model, which is a simplified version of a finite-volume numerical model To improve efficiency of our work, we propose using statistical models to approximate the Power Vs Location model outputs. In our experiment, Bayseian Additive Regression Trees (BART) model, outperforms traditional stationary Gaussian Processes (GP) model, and also the non stationary Tree Gaussian Processes (TGP) model, in predicting different non stationary power surfaces. Thus we incorporate BART into sequential design to predict the deterministic computer model outputs, which is the primary contribution of this research. We conduct simulation studies and apply our algorithm to placing turbine fences one at a time and two a time. Reasonably good predictions are obtained in both cases. We discover that if 10 fences are placed sequentially into the Minas Channel, the total maximum extractable power in creases from 4:76 108W to 31:57 106 W, but power generated by each turbine decreases as more fences are added. In addition, our results indicate that simultaneously placing multiple fences results in almost the same power generation as placing fences on the same power generation as placing fences one at a time.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.
https://scholar.acadiau.ca/islandora/object/theses:729