Mesh adaptation framework for embedded boundary methods for computational fluid dynamics and fluid‐structure interaction

Abstract

Summary Embedded Boundary Methods (EBMs) are often preferred for the solution of Fluid‐Structure Interaction (FSI) problems because they are reliable for large structural motions/deformations and topological changes. For viscous flow problems, however, they do not track the boundary layers that form around embedded obstacles and therefore do not maintain them resolved. Hence, an Adaptive Mesh Refinement (AMR) framework for EBMs is proposed in this paper. It is based on computing the distance from an edge of the embedding computational fluid dynamics mesh to the nearest embedded discrete surface and on satisfying the y+ requirements. It is also equipped with a Hessian‐based criterion for resolving flow features such as shocks, vortices, and wakes and with load balancing for achieving parallel efficiency. It performs mesh refinement using a parallel version of the newest vertex bisection method to maintain mesh conformity. Hence, while it is sufficiently comprehensive to support many discretization methods, it is particularly attractive for vertex‐centered finite volume schemes where dual cells tend to complicate the mesh adaptation process. Using the EBM known as FIVER, this AMR framework is verified for several academic FSI problems. Its potential for realistic FSI applications is also demonstrated with the simulation of a challenging supersonic parachute inflation dynamics problem.

Modeling, Simulation and Validation of Supersonic Parachute Inflation Dynamics during Mars Landing

“…The inflow conditions are the same as for Scenario 1. The background computational fluid domain is initially discretized by a mesh composed of Kuhn simplices [31,32]. This initial tetrahedral mesh contains 2, 778, 867 vertices and 16, 308, 672 tetrahedra.…”

An embedded boundary approach for resolving the contribution of cable subsystems to fully coupled fluid‐structure interaction

“…Therefore, adaptive mesh refinement (AMR) [55,56,57,58,59] is required. FIVER is equipped with a local refinement and coarsening algorithm based on the newest vertex bisection (NVB) [56,57,25], which enables the boundary layer and flow features to be efficiently tracked using the wall distance and a Hessian error indicator, respectively. However, for the cable subsystem, fully resolving its boundary layer is generally unaffordable, especially when the cable has large length-to-diameter ratio.…”

“…However, for the cable subsystem, fully resolving its boundary layer is generally unaffordable, especially when the cable has large length-to-diameter ratio. To obtain a minimally acceptable resolution we propose a new criterion for marking edges for refinement which constitutes a lightweight alternative to the wall distance criterion proposed in [25]. Specifically, when an edge of the fluid mesh is intersected twice by the cable's outer surface -which indicates that the cable is underresolved in this proximity -the edge is selected for refinement and subsequently bisected.…”

“…before adaption) discretized by 67,686 tetrahedra. To resolve the hose geometry, boundary layer and vortices, the fluid mesh is adaptively refined and coarsened by the newest vertex bisection algo-rithm [57,25]. The distance-based and Hessian-based mesh adaptation criteria are applied for boundary layer refinement and vortex refinement separately.…”

“…Both the "dressing" approach and the alternative master/slave kinematic approach are incorporated into the in-house Eulerian computational framework FIVER [21,22,23,24] -Finite Volume method with Exact two-material Riemann Problems. To track the boundary layers around the cable subsystem, adaptive mesh refinement [25] is applied to ensure reasonable mesh resolution in the vicinity of the cable subsystem.…”