A close look at different model types used by Syroco Live

The digital twins used by the Syroco platform are built through the assembly of models that represent the behaviour of each of the vessel's components. This post provides a (partial) inventory of the different types of models and their characteristics.

Physics-induced models

The physics-induced models are based on the state of the art of naval architecture. They model in a 3DoF simulator the behaviour of the ship in any kind of weather and speed conditions.

Calmwater model: the ship performance in calmwater is computed using CFD RANS computations including all the operational drafts, the speed profile of the ship and the leeway angles.

Wave added resistance model: the wave added resistance is computed with the SNNM model recommended by the ITTC, taking into account total sea wave height, period and direction, the STW and the specific geometry of the ship.

Windage model: the windage is computed using CFD RANS computations if the superstructure is available in 3D. Otherwise a parametric model based on the general arrangement is used.

Propeller model: the specific Kt, Kq, η curves are used if available (sea trial, towing tank or manufacturer). Otherwise the Wageningen parametric propeller model is used.

Rudder and appendage models: a parametric lifting surface model using state of the art formula is implemented. Specific studies using CFD analysis or data from the system provider can also be implemented.

Engine model: the specific SFOC curve of the engine evaluated during the engine factory test is used to convert brake power into fuel consumption. LNG engines are also available in the platform, including the boil-off model depending on the operational profile of the ship. Electric engines are also supported.

Fouling model: an empirical model is used for the initialisation of added resistance due to fouling. This added resistance is updated with data collected on the ship and takes into account the latest dry dock date, the time at sea and the time in port.

Data-driven models

Data-driven models are purely based on the data collected aboard, without any assumption of the physics.

Different types of data are collected: AIS data giving position and speed every 15 minutes, or onboard IoT data providing high-frequency sensor information (torque, fuel consumption, RPM, propeller pitch, etc). For each measurement point, hindcast weather is fetched to provide correlation between measurements and weather conditions.

These models give insight about the actual conditions the vessel has encountered in the past and provide information about how the ship is handled by the crew in certain conditions.

Syroco hybrid models

Syroco hybrid models are created using machine learning algorithms, taking as inputs data from both physics-induced models and data-driven models.

The hybrid models excel in predicting the ship behaviour in every kind of conditions. Their predictions are more comprehensive than data-driven models that are only relevant to accurately predict the behaviour of the ship under commonly encountered conditions.

Combining these approaches enables Syroco to confidently predict the ship's response in any circumstance, and also to build a digital twin of the ship even with limited data access or availability.