Characteristic-based and interface-sharpening algorithm for high-order simulations of immiscible compressible multi-material flows

详细信息    查看全文

• 作者：Zhiwei He ; Baolin Tian ; ^{tian_baolin@iapcm.ac.cn} ; Yousheng Zhang ; Fujie Gao
• 关键词：Multi-material flow ; Mie&ndash ; Gr&uuml ; neisen-type equation of state ; High-order ; Roe approximation ; Interface sharpening
• 刊名：Journal of Computational Physics
• 出版年：2017
• 出版时间：15 March 2017
• 年：2017
• 卷：333
• 期：Complete
• 页码：247-268
• 全文大小：3419 K
• 卷排序：333

文摘

The present work focuses on the simulation of immiscible compressible multi-material flows with the Mie–Grüneisen-type equation of state governed by the non-conservative five-equation model [1]. Although low-order single fluid schemes have already been adopted to provide some feasible results, the application of high-order schemes (introducing relatively small numerical dissipation) to these flows may lead to results with severe numerical oscillations. Consequently, attempts to apply any interface-sharpening techniques to stop the progressively more severe smearing interfaces for a longer simulation time may result in an overshoot increase and in some cases convergence to a non-physical solution occurs. This study proposes a characteristic-based interface-sharpening algorithm for performing high-order simulations of such flows by deriving a pressure-equilibrium-consistent intermediate state (augmented with approximations of pressure derivatives) for local characteristic variable reconstruction and constructing a general framework for interface sharpening. First, by imposing a weak form of the jump condition for the non-conservative five-equation model, we analytically derive an intermediate state with pressure derivatives treated as additional parameters of the linearization procedure. Based on this intermediate state, any well-established high-order reconstruction technique can be employed to provide the state at each cell edge. Second, by designing another state with only different reconstructed values of the interface function at each cell edge, the advection term in the equation of the interface function is discretized twice using any common algorithm. The difference between the two discretizations is employed consistently for interface compression, yielding a general framework for interface sharpening. Coupled with the fifth-order improved accurate monotonicity-preserving scheme [2] for local characteristic variable reconstruction and the tangent of hyperbola for the interface capturing scheme [3] for designing other reconstructed values of the interface function, the present algorithm is examined using some typical tests, with the Mie–Grüneisen-type equation of state used for characterizing the materials of interest in both one- and two-dimensional spaces. The results of these tests verify the effectiveness of the present algorithm: essentially non-oscillatory and interface-sharpened results are obtained.