Prospects of wind power prediction and variable operation in optimizing wind-powered reverse osmosis operation

Abstract

Reverse Osmosis (RO) is a dominant process in the desalination industry. However, concerns have been raised regarding its impact on the environment due to the dependency of commercial-scale plants on fossil fuels. Renewable Energy (RE) has been used in several studies to operate RO plants and decarbonize water production. However, the technology is either limited to small-scale plants, or large-scale plants that rely on a grid connection to meet the required water demand. This study is part of an international collaboration that aims to efficiently operate large-scale RO plants using wind energy and achieve the transition to fully sustainable RO. Variable-speed operation and modular operation are defined as strategies to operate the RO plant according to the available wind energy. As an initial step, this paper studies the combination of variable operation and wind-speed prediction in optimizing the operation of wind-powered RO. An Artificial Neural Network (ANN) was trained using a full year wind speed time-series to predict hourly average wind speed using a previously recorded 12-hour time series. The prediction exhibited high accuracy based on the regression analysis and would be implemented in the RO control system during the next stages of development. Wind speed prediction presents great potential for scheduling the startup/shutdown cycles of RO trains during modular operation. Currently, a pilot RO plant is being developed at Aston University for future testing. It is designed to deliver comparable operation to commercial RO plants by using commercial components arranged in a split feed flow configuration.