A Machine Learning based assessment of OTEC potential in the Bay of Bengal and it's harvesting approach

Abstract

Bangladesh has been steeply growing towards a concerning power demand. The coastal region and nearby islands surrounding the Bay of Bengal barely gets more than half of its required power through conventional power grid sources. This research attempts to look at the ocean thermal reserves of the Bay of Bengal and place a model 100 MW Net/150 MW gross OTEC power plant that could potentially alleviate the power shortage in the focused region. The potential sites were identified by analyzing big data obtained from HYCOM+NCODA using advanced data analytics. Placing the hypothetical plant, an estimated power output is calculated for different seasons. Cost estimations are found from an established model to provide an insight of the approximate levelized cost of electricity. A machine learning algorithm is trained to forecast the behavior of the temperature difference over the coming years to verify the consistency of the plant’s power output during its lifetime. Findings reveal that the estimated power ranges between 133 MW to 158 MW net power throughout the year. Levelized cost of electricity ranges 0.164 to 0.605 $/kWh for its low-medium-high estimates. Time series forecasting of historical data provide a clear image of the temperature difference in the coming years, indicating satisfactory consistency in generated power throughout the plant’s life. The work is novel as no prior work is noticeable in the OTEC field incorporates the use of modern data analytic tools or machine learning algorithms. Very few works in the field present optimum site selection as well as pertaining computations of the model plant’s performance. Even further novelty of the work can be posited to no comprehensive OTEC based research can be seen based on the Bay of Bengal or Bangladesh. The findings present an attractive argument and show promise, however, practical implementation of such a large plant requires multiple rounds of trial and errors to minimize failure and cost overload. Future research could implement a physical scaled model of the plant to observe its harvested power and associated costs. With modern advances in technology and possible government sanctions, the prospect of looking into OTEC as a renewable and environment friendly power alternative takes immediate priority.