Session: 06-11-01 Ocean Engineering Technology I

Paper Number: 78714

78714 - "Digital Twin of a Maneuvering Ship: Real-Time Estimation of Drift And Resistance Coefficients Based on Ship Motion and Rudder and Propeller Commands" 

The present work tackles the modeling problem concerning the initial steps of a Digital Twin (DT) application in a maneuvering ship. To envision a real implementation, further problems need to be solved, such as architecture design, communication network, digitalization purpose, etc., subjects which will not be aborded here. Instead, we concentrated on developing a well-rounded general system model that will enable the future employment of DT technology. Some examples can include performance analysis due to degradation of the hull, rudder, or thruster; decision support for maintenance scheduling; or even distance monitoring.

The paper focuses on real-time estimation of the main hydrodynamic coefficients, namely the longitudinal linear and quadratic resistance coefficients and linear maneuvering derivatives. The input to the system is the ship motion (obtained from GPS and Gyrocompass) and the commands to the propeller and rudder. Compared to the previous (OMAE2021-62899), we included the propeller and rudder models. The rudder and propeller forces and moment were considered fully known in the previous work, which is a difficult task in practical terms. The new proposed system model requires the thruster rotation and the rudder angle – information much easier to access and highly accurate to obtain.

Another improvement concerns the parameter estimation methodology. We used the Unscented Kalman Filter (UKF) instead of the Extended Kalman Filter (EKF). For nonlinear dynamics, this change eased the calculation burden of obtaining the unknown model parameters.

The proposed real-time system identification system architecture is based on 3 degrees of freedom maneuvering model of a single rudder and single propeller ship. The method was tested in a simulated environment based on PyDyna simulator (a ship maneuvering simulator implemented on python based on the mathematical model adopted in the TPN-USP Ship Maneuvering Simulation Center). Data from motion sensors were mimicked by inducing a Gaussian white noise in the data retrieved from the simulator in real-time intending to better represent a real-world scenario.

Some standard maneuvers were tested, revealing this method to be easier integrated with already well-established tests procedures or even real navigation. Besides its relatively quickly computational cost, it possibly presents as a convenient preliminary parameter assessment until more precise and time-consuming methods such as CFD are evoked.

Presenting Author: Humberto Akira Uehara Sasaki Universidade de São Paulo / Numerical Offshore Tank (TPN-USP)

Authors:

Humberto Akira Uehara Sasaki Universidade de São Paulo / Numerical Offshore Tank (TPN-USP)
Pedro Cardozo De Mello Universidade de São Paulo / Numerical Offshore Tank (TPN-USP)
Eduardo Aoun Tannuri Universidade de São Paulo / Numerical Offshore Tank (TPN-USP)