Iterative 3D layout optimization and parametric trade study for a reconfigurable ocean current turbine array using Bayesian Optimization

doi:10.1016/j.renene.2018.05.040

Return

Iterative 3D layout optimization and parametric trade study for a reconfigurable ocean current turbine array using Bayesian Optimization

2018-11-01

PRE

Ali Baheri *

Praveen Ramaprabhu

Christopher Vermillion

DOI：10.1016/j.renene.2018.05.040

Original

Original request for help

Save

Abstract

En 中文

In this paper, we present an online approach for optimizing the 3D layout of a reconfigurable ocean current turbine (OCT) array. Unlike towered turbines, most OCT concepts for Gulf Stream energy harvesting involve tethered systems. The replacement of towers with tethers provides the opportunity for OCTs to adjust their locations within some domain by paying out/in tether to adjust depth and manipulating control surfaces (elevators and rudders) to adjust longitudinal and lateral positions. The ability to adjust the OCT positions online provides the capacity to reconfigure the array layout in response to changing flow conditions; however, successful online array layout reconfiguration requires optimization schemes that are not only effective but also enable fast convergence to the optimal configuration. To address the above needs, we present a reconfigurable layout optimization algorithm with two novel features. First, we describe the location of each turbine through a small set of basis parameters; the number of basis parameters does not grow with increasing array size, thereby leading to an optimization that is not only computationally tractable but is also highly scalable. Secondly, we use Bayesian Optimization to optimize these basis parameters. Bayesian Optimization is a very powerful iterative optimization technique that, at every iteration, fuses a best-guess model of a complex function (array power as a function of basis parameters, in our case) with a characterization of the model uncertainty in order to determine the next evaluation point. Using a low-order analytical wake interaction model, we demonstrate the effectiveness of the proposed optimization approach for various array sizes. (C) 2018 Elsevier Ltd. All rights reserved.

Keywords:

Bayesian Optimization

Layout optimization

Ocean current turbine array

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.