Interface-preserving level set method for simulating dam-break flows

Abstract

An interface-preserving level set method that solves advection and re-initialization equations for simulating three-dimensional dam-break flows is developed. This method solves mass transfer problems on a uniform staggered Cartesian grid. The advection equation that is used to advect the level set function for capturing the interface is discretized by a proposed fourth-order spatial discretization scheme. This scheme is dispersion-relation-preserving and is compact-reconstruction weighted essentially non-oscillatory (DRP-CRWENO4). This scheme is compared with a previous fifth-order, weighted, essentially non-oscillatory (WENO5) scheme and can represent an interface more accurately, while exactly preserving mass conservation. This level set approach introduces a mass correction term into the re-initialization equation based on local interface-preserving conditions. An explicit Adams–Bashforth algorithm on a staggered Eulerian grid is used for the Navier–Stokes solver. The point successive over-relaxation method is then employed to solve the resulting linear system. Two one-dimensional wave propagation problems are simulated to verify the proposed DRP-CRWENO4 scheme, which is shown to be capable of effectively capturing large gradients with fourth-order accuracy. To demonstrate their resolution, the two advection schemes (WENO5 and DRP-CRWENO4) are applied in two two-dimensional benchmark cases, i.e., a vortex deforming problem and Zalesak's disk problem, where simulation results of both schemes are compared against each other. Demonstration study is then further extended to three-dimensional cases of the vortex deforming problem and Zalesak's sphere problem, and simulation results agree well with those using hybrid particle level set method. Finally, several dam-break problems with and without obstacles are investigated to validate the coupled two-phase incompressible flow and level set method solver. The results for the predicted flow structure and mass conservation properties are compared with the reported experimental data or numerical simulations.