Optimal Control Energy Management Strategy for Hybrid Propulsion and Power Supply Vessels using Data-Driven Load Forecasting

Optimal Control Energy Management Strategy for Hybrid Propulsion and Power Supply Vessels using Data-Driven Load Forecasting

Slagter, Simeon (TU Delft Mechanical, Maritime and Materials Engineering; TU Delft Transport Engineering and Logistics)

Coraddu, A. (mentor)
Kalikatzarakis, Miltos (graduation committee)
Polinder, H. (graduation committee)
Kopka, T. (graduation committee)

Delft University of Technology

Mechanical Engineering | Multi-Machine Engineering

2023-02-20

Hybrid technology can significantly reduce fuel consumption and emission for vessels that have high power demand peaks followed by long periods of low loading. Hybrid technology refers to powertrain layouts that consist of hybrid propulsion and/or hybrid power supply. Advanced energy management strategies (EMS) are required to make optimal use of these available power resources. In this paper it is investigated how much fuel consumption reduction can be achieved by applying a causal, real-time equivalent consumption minimization strategy (ECMS) to a hybrid propulsion and hybrid power supply plant with a power load forecasting scheme, for a case study vessel: The Holland-class offshore patrol vessel. The forecasting tools evaluated are Linear regression, moving average, ARIMA and recurrent neural networks (RNN). The RNN outperformed the other methods and is able to predict the power demand for up to 48 seconds, while maintaining a mean absolute percentage error of under 5%. An optimization-based controller is combined with an ECMS approach which assigns an equivalent consumption cost to the battery. The controller is able to identify the power split for the hybrid propulsive system, and the power split for the hybrid power supply. A simulation proved that 0.77% fuel savings are achieved with a 400 kW battery, compared to a no-battery scenario. Fuel savings could not be proven for the EMS with a control horizon of 48 seconds, leveraging power predictions supplied by the RNN, due to limiting factors. The limiting factors are the combination of the small control horizon, the limited battery capacity compared to the overall power demand and the limited tuning of the ESFC curve.

Energy management system
Propulsion control
Hybrid vehicles
Marine systems
Hybrid energy supply

http://resolver.tudelft.nl/uuid:54ab548a-2b1a-407f-8033-8e8766c3e742

2025-02-10

Student theses

master thesis

© 2023 Simeon Slagter