Digital twins, at the core of Syroco EfficientShip, are built through the assembly of models that represent the different parts of the ship: hull, rudder, propulsion, superstructure, etc. Each model computes the physical forces applied to the ship part, in a given configuration and under specific environmental conditions (wind speed and direction, wave height and frequency, current, etc.).
Loads that apply to the part are modelled using different types of input: CFD studies, towing tank experiments, on-board IoT data from sensors processed with machine learning, and parametric models. Let’s dive into parametric models.
A parametric model is built around an explicit formula that gives a force intensity for a set of parameters. Such models originate from a theoretical modelisation (similar to the gravity force from Newton’s theory: F=m.g) or from an empirical study (like Holtrop & Mennen to compute ship water resistance).
EfficientShip lets the user create (or use existing) parametric models for all the forces encountered on a ship at sea:
- Calmwater hydrodynamic resistance
- Wave added resistance
- Engine and propeller propulsion system
- Wind assisted ship propulsion systems (sails, kites, wings, rotors, etc.)
- Rudder and appendices
- Superstructure windage
Data used to create these models is typically made of physical characteristics of the component. To list a few examples:
- For the hull, the model will be generated from the overall geometry parameters: length, beam, draught, shape type, etc.
- For a rudder or appendix, input data includes surface, aspect ratio, cord, angle of attack, etc.
- For the propulsion chain, propeller laws and engine laws are used to model the relationship between thrust at the propeller and fuel consumption at the engine intake.
Depending on the complexity of the geometry and the dynamics of the component, required inputs may range from half a dozen to twenty or more parameters.
Because they are built around an explicit formula with only few parameters, parametric models are usually very fast and versatile, and they are much quicker (and therefore less expensive) to build and run than object-specific models that use CFD, experimental or IoT data. However, this also carries drawbacks: parametric models are less accurate than the object-specific models and are therefore considered low-fidelity models.
In EfficientShip, parametric models are often used to allow fast and broad pre-design exploration of a large field of ships and solution. They allow clients to try or explore new features on a ship before starting expensive and extensive design studies. They are also a base on which higher fidelity models are built, using data and model interpolation to enrich the low fidelity models through iterative process.