Analysis, prediction and variation of the hydrodynamic properties of wind-assisted ships
Moritz Konrad Schäfer, Technische Universität Berlin
Wind-assisted ship propulsion (WASP) is among the most promising technologies to significantly reduce greenhouse gas (GHG) emissions from maritime transport and to meet the climate targets set by the International Maritime Organization (IMO). In both new designs and retrofit projects involving wind propulsion, it is essential to rapidly and accurately predict the hydrodynamic behavior of the hull and the effect of form variations.
In contrast to traditional propulsion, wind propulsion introduces large lateral forces that affect not only resistance but also the course keeping ability, i.e. added drag due to drift, and yaw stability—quantified by the center of lateral resistance (fgh). Accurately capturing these effects is crucial for performance prediction of wind-assisted ships.
Traditional empirical maneuvering models are commonly used to estimate side forces and drift-induced drag. However, their accuracy is insufficient for making reliable design decisions, as these models rely on only a few geometric parameters and fail to capture variations in hull form. To address this limitation, maneuvering coefficients can be derived from Computational Fluid Dynamics (CFD) simulations or model tests by fitting polynomials to the resulting forces and moments. Although this method is accurate, it is computationally or economically expensive and therefore its practicality in early design stages is limited.
This thesis, as part of the EU project Retrofit 55, aims to develop an efficient and accurate methodology for predicting hydrodynamic forces and moments on drifting hulls. Using both potential flow and viscous RANS CFD tools, the hydrodynamic loads are analyzed and de- composed into potential and vortex lift, crosflow drag (and its center), and the Munk moment, based on studies of the KVLCC2 and a Kamsarmax bulker.
A structured workflow is proposed, consisting of three levels of fidelity:
• Level 1: A purely surrogate-based model using artificial neural networks (ANNs) trained on CFD data from a range of hull variants (e.g., different main dimensions and general form features) to predict side force coefficients, added drag, and yaw stability.
• Level 2: A hybrid approach combining the surrogate model with a single CFD simula- tion to improve accuracy while retaining computational efficiency.
• Level 3: The highest-accuracy method, involving two viscous CFD simulations and one potential flow simulation to fully resolve and decompose hydrodynamic forces and moments.
The proposed approach supports design-stage decision-making by balancing accuracy and computational cost
Results demonstrate that the lift and drag of a drifting hull can be accurately estimated by decomposing the forces into potential lift, vortex lift, and crossflow drag. Additionally, potential flow calculations effectively predict the Munk moment across all drift angles using a single simulation, as the added mass coefficient (M₃₃ – M₁₁) remains constant. Consequently, all three fidelity levels provide high accuracy in predicting drift behavior under external lateral forces.
