High-order compact finite volume schemes for solving the Reynolds averaged Navier-Stokes equations on the unstructured mixed grids with a large aspect ratio
Introduction
High-order methods have shown great capability in simulating multi-scale flow problems such as turbulences [1]. For flows with complex geometries, high-order numerical methods on the unstructured grids are preferred. Recent advances have been made in the finite volume (FV) method [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], discontinuous Galerkin (DG) methods [13], [14], [15], [16], spectral volume (SV) [17]/spectral difference (SD) methods [18], PnPm procedures [19], FR [20]/CPR [21] methods, hybrid FV/DG methods [22], [23], [24], and multi-moment methods [25] for solving multi-scale flow problems on the unstructured grids.
An active area of the study of the high-order schemes on the unstructured grids is the simulation of the turbulent flows governed by the Reynolds averaged Navier-Stokes (RANS) equations. The current design tools in aircraft industries are primarily based on the RANS equations [26]. RANS equations are also necessary for the wall modeling of the large eddy simulation (LES) [27], as well as the hybrid RANS/LES modeling such as the detached eddy simulation (DES) [28]. Various turbulent models have been proposed [29] to meet the needs of industrial applications. Since it is widely accepted that high-order schemes outperform their second-order counterparts in achieving grid-independent solutions and in resolving important flow structures such as the turbulent boundary layers, it is necessary to develop high-order accurate RANS solvers. To this end, high-order FV [31], [32], DG [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], SD [43], FR/CPR [44], [45], and the hybrid FV/DG methods [46] have been applied to solve RANS equations in recent studies. The Spalart Allmaras (SA) one-equation model [30] is widely used among the turbulent models due to its good performance and easy implementation. In this paper, we will focus our discussion on the SA model.
It is still challenging to solve the turbulence transport equations such as the SA model using high-order numerical schemes. The first issue arises from the large aspect ratio of the near-wall grids to resolve turbulent boundary layers [35]. The time-step of the explicit time-marching scheme is significantly reduced due to the small grid size in the wall normal direction. The implicit temporal discretization scheme is preferred to improve the solution efficiency, as suggested in [35]. When using high-order FV schemes, it has also been reported that the grid aspect ratios affect the reconstruction accuracy [32]. In [32], a reconstruction scheme based on local curvilinear normal-tangential coordinates was proposed to enhance the reconstruction accuracy on grids with large aspect ratios. The second issue is the stiff source terms in the turbulence modeling equations. Negative turbulent viscosity in the outer edge region of the boundary layer can be predicted [32], [33] due to the stiffness of the turbulence model. To suppress the appearance of negative turbulent viscosity, the standard SA model is suggested to be replaced by the negative SA model [54], which has been proven to be effective [32].
In the present paper, we will study the numerical simulation of the RANS equations closed by the SA model using the recently proposed high-order compact FV schemes based on the variational reconstructions [4] on the unstructured mixed grids. A distinctive advantage of these schemes is that all operations of the algorithms are based on a compact stencil, and the large stencil problem associated with traditional high-order finite volume schemes is resolved entirely. Another advantage is that the variational reconstructions are non-singular with unique solutions.
Two outstanding problems for applying the variational reconstruction FV schemes to solve the RANS equations closed by the SA model are addressed in the present paper. The first one is to design an effective reconstruction algorithm for the quadrilateral grids with a large aspect ratio. The present FV scheme can be applied on the mixed grids with triangular and quadrilateral control volumes. However, near the solid walls of the turbulent boundary layers, the preferred meshes are quadrilateral. Therefore, optimization of the FV scheme for the quadrilateral grids with a large aspect ratio will be studied. The variational reconstruction relies on minimizing a particular functional with adjustable parameters. It is therefore possible to optimize these parameters to improve the performance of the FV scheme. For high-order accurate reconstructions, it turns out that the optimization procedure is challenging. To simplify the analysis, we propose a three-step optimization procedure, including 1) the derivative weights optimization, 2) the geometric weights optimization, and 3) the generalization to general unstructured grids. According to the tests based on a manufactured solution of physical importance, the optimized scheme significantly improves the reconstruction accuracy on the quadrilateral grids with large aspect ratios.
The second one is the avoidance of the possible negative turbulent viscosity in the solutions of the SA model. Different from other methods for eliminating the negative turbulent viscosity mentioned previously in this section, a new procedure called the exponential decay procedure (EDP) is proposed in the present paper. By analyzing the asymptotic behavior of the SA model when the turbulent viscosity is close to zero, we find that the turbulent viscosity will undergo an approximately exponential decay. This physics-based reasoning leads to EDP, a numerical positivity preserving procedure. The EDP is straightforward and effective and can be easily coupled with high-order spatial discretization and implicit time integration schemes. Furthermore, through EDP, the standard SA model is sufficient to achieve the positivity preserving solution, and no other modification such as the use of the negative SA model is needed.
Several test cases are solved numerically with the proposed FV schemes and the EDP. Numerical tests show significant advantages of the high-order schemes in predicting the skin frictions, capturing some important flow structures, and achieving grid-independent solutions. The numerical simulations also indicate that the proposed schemes are sufficiently robust for practical applications.
Section snippets
Governing equations and boundary conditions
The two-dimensional, compressible conservative RANS equations in the non-dimensional forms are where In Eqs. (1) and (2), is the vector of the conservative variables, represent the convective fluxes, and denote the viscous fluxes. Pressure is given by the equation of state for the ideal gas where the gas constant is with the specific
Numerical tests
In this section, we will solve some benchmark test cases to verify the performance of the proposed high-order FV schemes in solving the RANS equations closed by the SA model. The reconstruction is optimized for large aspect ratio grids, and the EDP-based algorithm for preserving turbulent viscosity positivity is used. In addition, we will discuss whether the high-order FV scheme is superior to the low-order one in improving the prediction accuracy, capturing critical flow structures and
Conclusions
The large aspect ratio grid and the presence of negative turbulent viscosity bring challenges in the practical application of the high-order numerical schemes for solving complicated turbulent flows. In this paper, by optimizing the reconstruction scheme on grid with large aspect ratio and eliminating the negative turbulent viscosity through the EDP, an accurate and robust high-order FV solver on the unstructured mixed grids solving RANS equations closed by the SA model has been developed. The
CRediT authorship contribution statement
Qian-Min Huang: Methodology, Coding, Computing, Writing – original draft preparation. Yu-Xin Ren: Conceptualization, Supervision, Writing – reviewing and editing. Qian Wang: Methodology. Jian-Hua Pan: Methodology.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgement
This work is supported by National Natural Science Foundation of China (92152201), the China Postdoctoral Science Foundation (2019M660613) and the National Numerical Wind Tunnel Project.
References (70)
- et al.
High-order CFD methods: current status and perspective
Int. J. Numer. Methods Fluids
(2013) - et al.
Compact high order finite volume method on unstructured grids I: basic formulations and one-dimensional schemes
J. Comput. Phys.
(2016) - et al.
Compact high order finite volume method on unstructured grids II: extension to two-dimensional Euler equations
J. Comput. Phys.
(2016) - et al.
Compact high order finite volume method on unstructured grids III: variational reconstruction
J. Comput. Phys.
(2017) - et al.
High-order compact finite volume methods on unstructured grids with adaptive mesh refinement for solving inviscid and viscous flows
Chin. J. Aeronaut.
(2018) - et al.
Compact high order finite volume method on unstructured grids IV: explicit multi-step reconstruction schemes on compact stencil
J. Comput. Phys.
(2019) - et al.
Higher order solution of the Euler equations on unstructured grids using quadratic reconstruction
(1990) - et al.
Quadratic reconstruction finite volume schemes on 3D arbitrary unstructured polyhedral grids
(1999) Quasi-ENO schemes for unstructured meshes based on unlimited data-dependent least-squares reconstruction
J. Comput. Phys.
(1997)Weighted essentially non-oscillatory schemes for the interpolation of mean values on unstructured grids
J. Comput. Phys.
(1998)