Analysis of Coupling Techniques for Overset-Grid Finite-Volume Methods
Produktform: Buch / Einband - flex.(Paperback)
Multi-body hydrodynamics featuring mutual interaction and contact between rigid or
flexible floating bodies is an active area of research for complex marine engineering
devices. An example refers to offshore supply vessels, where it is important to consider
operations in close proximity of other vessels or structures, e.g. the landing manoeuvre
of the vessel at an offshore foundation (Luo-Theilen and Rung 2017). Other examples,
which are subject to significant hydrodynamic interaction forces, are the simulation
of ship collisions (Rudan and Volari´c 2016) and the collision between ice floes and an
ice-cruising vessel (Janßen et al. 2017).
Particularly due to safety reasons, it is essential to know and to understand the flow
field around moving bodies, the arising hydrodynamic forces and the resulting body
behaviour within decisive marine operations already in the design process. While not all
aspects can be captured economically within analytical or experimental examinations,
numerical simulations are meanwhile established as a reliable design tool. Due to the
continuous technological progress of numerical hard- and software, computational fluid
dynamics (CFD) nowadays provides the opportunity to investigate complex marine
procedures prior to critical incidences. Especially viscous CFD methods give an extensive
insight into the occurring physical phenomena. Industrial viscous flow simulations
often follow a finite-volume approach. However, fluid dynamic simulations of multiple
floating bodies, which feature large relative motion, are still challenging using these
grid-based Eulerian methods. In this regard, overset-grid methods offer a versatile
approach. They simplify the grid generation by using modular grid components and at
the same time often improve the grid quality. The technique comes at the expense of (a)
an elaborate priority management and cell blanking for regions covered by multiple
grids, (b) a challenging interpolation-based coupling between disjunct grids and (c)
complex dynamic load balancing efforts for parallel applications. All these algorithmic
challenges require efficient search algorithms used to manage the grid connectivity
(Hadži´c 2006, Löhner 2008, Brunswig and Rung 2013).
Many industrial CFD applications are based on unstructured-grid finite-volume
methods and employ a co-located, cell-centred variable arrangement. A common
implicit approach to couple multiple unstructured three-dimensional grids in a
cell-centred scheme is to interpolate neighbouring field values onto the partner grids,
which creates an implicit link between the grids on the level of the equation system.
Since an interpolation-based coupling procedure is restricted to local information, it is
insufficient to guarantee that the global sum of the mass fluxes across the overlapping
interfaces vanishes. A simple example refers to the exterior surface of a foreground
1
1 Introduction
grid, which is fully embedded in the interior of a background grid, cf. the schematic
representation on the left-hand side of Figure 1.1. The arrangement implies that the
sum of the mass fluxes along the exterior boundary of the foreground grid must vanish,
which cannot be guaranteed by an interpolation-based grid coupling without dedicated
corrections. Consequently, such overset-grid approaches violate the inherent mass
conservation of finite-volume methods. This issue is of significance for both the global
mass balance of the entire simulation as well as the local balance of each domain. In this
context, the blanked area within the background domain yields additional boundary
fluxes (displayed on the right-hand side of Figure 1.1) which also contribute to the
mass balance. Due to the fact that incompressible finite-volume methods directly use
the mass defect when solving for the pressure, severe pressure fluctuations can be
provoked.
Figureweiterlesen
50,00 € inkl. MwSt.
kostenloser Versand
lieferbar - Lieferzeit 10-15 Werktage
zurück