A trip-ahead strategy for optimal energy dispatch in ship power systems

Optimizing ship power systems with diesel generators such as the platform supply vessel (PSV) has become a pressing issue due to the emission of carbon dioxide. This paper investigates the optimal operation of diesel generators in marine power systems, particularly the PSV. The investigation is mainly focused on carbon dioxide emission and fuel consumption in a ship's mission. In this regard, this paper presents a clear optimization strategy, called a trip-ahead to determine the best operation schedule of generators to supply the electricity demand for the next days of a PSV. The PSV has six generators (i.e., four primary and two auxiliary diesel generators) with two different fuel consumption curves and carbon dioxide emission. The trip-ahead algorithm's objective functions consist of minimization of cost and emission during the complete PSV mission. A power level is assigned to each generator for each hour of the next day of a 5-day trip, taking into account the fuel consumption per kilowatt-hour (kWh). The prepared sample of the load profile is the historical demand records of a real PSV. Additionally, this paper compares the results of the proposed approach with other optimization algorithms such as the Genetic algorithm (GA), Particle Swarm Optimization algorithm (PSO), and the software HOMER Pro optimization tool. Moreover, this paper presents the sensitivity analysis to compensate for possible errors in prediction demand for the next trip. The results prove the proposed algorithm in comparing GA and PSO is more accurate and the calculation velocity of the proposed algorithm when objective functions are cost and emission is about 27% and 46% better than the PSO and about 36% and 62% better than the GA, respectively.

Automated summary

This paper investigates the optimal operation of diesel generators in marine power systems, particularly the PSV, and proposes a trip-ahead algorithm to determine the best operation schedule of generators for PSV. The results show that the proposed algorithm is more accurate when objective functions are cost and emission compared to genetic algorithm and particle swarm optimization algorithm.