Design Optimization of Battery-Electric Marine Vessels via Geometric Programming

This paper proposes formulating conceptual-stage vessel design optimization problems as geometric programs, which can be transformed into convex optimization problems. Convex optimization offers significant advantages in efficiency, reliability and automation potential over the general nonlinear optimization approach typically used in naval architecture. Focusing on battery-electric vessels, geometric program compatible models are derived for lithium-ion cells, power converters, propulsion motors and propellers. Preliminary hull form development, stability calculation and structural design are also presented in the context of geometric programming. The modeling approach is applied to study optimal battery sizing for a coastal bulk carrier sailing in varying operational conditions. Using open-source software tools, the battery sizing problem is solved in less than a second on a standard desktop computer. Local sensitivity information encoded by optimal dual variables reveals that increasing the cell discharge upper bound by 1% decreases the optimal total number of cells by more than 1%. On the other hand, the sensitivities of cell volume and maximum discharging current parameters are zero, indicating that the constraints involving these parameters do not govern the solution.

Automated summary

This paper proposes using geometric programs to optimize conceptual-stage vessel design, which is more efficient, reliable, and automated compared to traditional nonlinear optimization methods used in naval architecture. Specifically focused on battery-electric vessels, the study presents geometric program compatible models for different components, such as lithium-ion cells and power converters. The modeling approach is applied to calculate optimal battery sizing for a coastal bulk carrier, and the problem is solved in less than a second using open-source software tools on a standard desktop computer. Sensitivity analysis reveals the impact of specific parameters on the optimal solution.