Mesh-Free Simulation of Port Hydrodynamic Problems



Figure 1: Simulation of a bow thruster induced scouring close to quay wall. The jet is coloured according to the particle velocities.

Background and Objectives

The principal aim of the project is the development of a computational procedure to investigate port-hydrodynamic problems. The employed apporach is based on the Lagrangian  Smoothed-Particle-Hydrodynamics (SPH) method. We refer to our simulation code as GADGET-H2O, as it represents a modification of the cosmological particle code GADGET-2 [1].

Figure 1: Potholes generated by the ship's propulsion and manoeuvring devices.

Figure 2: Potholes generated by the ship's propulsion and manoeuvering devices.

Investigated examples refer to ship (un)berthing. Emphasis is given to the generation of potholes (scours), which are induced by the ship's propulsion and manoeuvering devices (e.g. transverse thrusters and propellers) and significantly weaken the quay-wall support. The application of the methodology strives for an assessment of potential counter-measures.

Computational Approach

Harbor flows are governed by confined- & shallow-water effects. The wetted domain dynamically changes during the maneuvering process. Moreover, the quay-walls and the harbor bed are exposed to severe loads of the ship's propulsion and maneuvering devices.


Figure 3: Particle approximation of a 230 m container vessel with starting propeller and bow thruster. The colouring of the water particles and vector arrows refer to fluid velocity.   

The predictive challenge refers to the occurence of multiple continua (fluid-soil) and multiple flow-phases (air-water) as well as the large relative motions to be considered in generally confined regimes. The latter makes grid-based Eulerian simulation methods (e.g. FV- or FE-methods) undesirable, due to the severe geometrical restrictions of a grid-based discretization approach.

Mesh-free Lagrangian methods, like Smoothed-Particle-Hydrodynamics, which discretize the fluid rather than the domain, are of advantage when large relative motions should be captured. The SPH-method is inherently unsteady and applicable to problems dealing with multiple phases and continua.

SPH provides an integral representation of the flow field. The fluid is modeled by a finite amount of particles particles, carrying a mass and several field-related properties, e.g. density, pressure, velocity, temperature or turbulence energy. To evaluate the local field properties, a integrated weighted average is performed for a set of neighboring particles using a smoothing kernel function. Accordingly, the spatial derivatives of the governing transport equation (continuity, momentum, energy etc.) are evaluated analytically. The resulting system of ordinary differential equations is advanced in time, using established explicit numerical-integration schemes, e.g. Crank-Nicholson or Runge-Kutta.



Efforts are concerned with the following aspects 

  1. viscous fluids and turbulent flows  

  2. multiple continua  

  3. prediction of pressure in incompressible media  

  4. CAD-related pre-processing tool  

  5. parallel performance


Parallel Performance 

Figure 4: Parallel performance.


[1] Springel, V. (2005), The cosmological simulation code GADGET-2, Mon. Not. R. Astron. Soc. 364, 1105–1134