Convex programming for optimal control of a fuel cell hybrid ferry

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Insinööritieteiden korkeakoulu | Master's thesis
Degree programme
Master's Programme in Mechanical Engineering (MEC)
85 + 10
The limits of global warming stated in the Paris Agreement and the resulting need to reduce greenhouse gas emissions as well as economic considerations drive the international shipping industry towards lowering vessel emissions. Additionally, PM2.5, SOx and NOx emissions need to be decreased due to scientifically based human health concerns. This is especially important in coastal regions, where short sea vessels operate. Therefore, this work focuses on a ferry powered by a fuel cell and a battery allowing for locally emission-free operation. A holistic optimization approach is chosen to achieve maximum fuel efficiency and to minimize the costs. The energy system, the fixed pitch propeller and the longitudinal dynamics of the vessel are modeled. The load sharing, the operating point of the propeller and the vessel speed are chosen as the main decision variables. The resulting continuous intertemporal optimal control problem traditionally is solved with algorithms based on dynamic programming. However, this approach suffers from the “curse of dimensionality” excluding the algorithms from real-time applications and offline design studies with the need to quickly run several hundreds of iterations. Thus, a more efficient alternative, convex optimization, is chosen. In a convex optimization framework only specific types of functions are allowed. Therefore, the main contribution of this work is to formulate a convex optimization model for the outlined purpose. Especially the convex propeller model and the convex vessel dynamics are to be highlighted. Another special feature is the model formulation in the spatial domain. This allows for the consideration of speed limits in the coastal areas. The potential of the model is analyzed and the necessity of considering longitudinal dynamics for short sea vessels is proven. The energy consumption with vessel dynamics is 3.7 % higher than without and the load profiles of fuel cell and battery differ significantly in the case study conducted. The sensitivity of the energy consumption regarding selected parameters is investigated and among these parameters the maximum trip time is found to have the most influence. To conclude, possible use cases like model predictive control and resulting future research opportunities are discussed.
Tammi, Kari
Thesis advisor
Ritari, Antti
optimal control, convex optimization, hybrid energy system, power management, fuel cell, maritime