Abstract
One of the crucial issues of computational fluid dynamics is how to discretize the viscous terms accurately. Recently, an attractive and viable alternative numerical method for solving the compressible Navier–Stokes equations is proposed. The first-order hyperbolic system (FOHS) with reconstructed discontinuous Galerkin (rDG) method was first proposed to solve advection–diffusion model equations and then extend to compressible Navier–Stokes equations. For the model advection–diffusion equation, the proposed method is reliable, accurate, efficient, and robust, benefiting from FOHS and rDG methods. To implement the method of compressible Navier–Stokes equations, the gradients of density, velocity, and temperature are introduced as auxiliary variables. Numerical experiments demonstrate that the developed HNS + rDG methods are able to achieve the designed order of accuracy for both primary variables and their gradients.
Keywords
- reconstructed discontinuous Galerkin
- hyperbolic Navier–Stokes
- first-order hyperbolic system
- integration methods
- laminar flow
1. Introduction
The compressible Navier–Stokes equations are useful because they describe the physics of many phenomena of scientific and engineering interests. They are widely used but not limited to the fields of fluid dynamics, aeronautical engineering, and aerodynamics engineering. The Navier–Stokes equations are certain partial differential equations that describe the motion of viscous fluid substances. The compressible Navier–Stokes equations mathematically express the continuity, the theorem of momentum, and the theorem of energy of Newtonian fluids. These equations arise from applying Isaac Newton’s second law to fluid motion, with the assumption that the stress in the fluid is the sum of a diffusing viscous term and a pressure term. The diffusing viscous term is proportional to the gradient of velocity. The difference between Euler equations and Navier–Stokes equations is that Euler equations are only valid for inviscid flow, while Navier–Stokes equations take viscosity into account. Euler equations are hyperbolic, while Navier–Stokes equations are parabolic.
Although the discontinuous Galerkin (DG) method is a natural choice for hyperbolic problems, it becomes far less certain when it comes to elliptic or parabolic problems. An alternative approach for viscous discretization, which reformulates the viscous terms as a first-order hyperbolic system (FOHS), was given in Refs. [1, 2]. Another FOHS formulation [3, 4] was developed by including the gradient quantities as additional variables. Over several years of development, the FOHS method has been shown to offer several distinguished advantages. First, the well-developed numerical schemes for hyperbolic systems can be directly applied to viscous problems. Second, it enables high-quality gradient prediction even on fully irregular grids. This feature shows its significance when it comes to viscous flow simulations on complex grids, where the qualities of the meshes are likely to be highly irregular. Third, due to the fact that the second-order derivatives in the governing partial differential equations (PDE) are replaced and eliminated, the developed scheme shows speedup and robustness for the iterative solvers. Finally, the FOHS method can improve viscous discretization as well as inviscid discretization. Due to these favorable characteristics, the hyperbolic methods have been implemented in various applications, including diffusion [5], anisotropic diffusion [6], advection–diffusion [7], Navier-Stokes (NS) equations [8], and three-dimensional compressible NS equations with proper handling of high Reynolds number boundary layer flows [8].
To make it easier for the readers to understand, the classical computation methods for fluid dynamics are first discussed, and the reconstructed discontinuous Galerkin (rDG) method is introduced. FOHS for linear advection–diffusion equation is then given. Eventually, the hyperbolic Navier–Stokes with rDG method is given.
2. Integration methods for first- and second-order operators
Computational fluid dynamics (CFD) is, in part, the art of replacing the governing partial differential equations of fluid flow with numbers and advancing these numbers in space and time to obtain a final numerical description of the flow [9]. To achieve this goal, accurate spatial and temporal discretization becomes necessary. CFD methods on unstructured grids mainly fall into three categories: finite element (FE) methods, finite volume (FV) methods, and DG methods.
2.1 Finite volume methods
Finite volume methods assume that the underlying solution is constant in each cell, and a Riemann problem [10] arises as discontinuity occurs at the boundary where two elements are adjacent to each other. For first-order operator
the finite volume method gives the formulation as
where
2.2 Finite element methods
Finite element methods are classically used for elliptic or parabolic problems. In the finite element method, a given domain is viewed as a collection of sub-domains, and over each sub-domain. Then, it becomes easier to represent a complicated function as a collection of simple polynomials [13]. As shown in Figure 1, the underlying solution is continuous at the element boundary. For second-order operator
FE formulation gives
where
2.3 Discontinuous Galerkin methods
DG methods combine the advantages of finite volume and finite element methods. While assuming the underlying solution to be polynomial on each element, the Riemann problem is solved at the element boundary. For the first-order operator (Eq. (1)), the DG formulation gives
In practice,
As the DG method assumes that the solution is distributed as a polynomial function on each element, the value of
2.4 Reconstructed discontinuous Galerkin methods
In rDG methods, a higher-order solution is reconstructed based on the underlying solution. (Eq. (5)) becomes
and (Eq. (6)) becomes
where
3. FOHS for linear advection: diffusion equation
The two-dimension linear advection–diffusion equation is written as
where
By adding pseudo time (
where
where
To simplify the mathematics, the advection term and the diffusive term are considered separately.
Consider the Jacobian of the flux projected along
where
where
The only nonzero eigenvalue of the advective Jacobian is
One can see that the diffusion progress can be regarded as a wave propagating isotropically according to the first two nonzero eigenvalues. As for the third eigenvalue, it indicates the inconsistency damping mode [18]. Clearly, the steady solution is independent of the free parameter, relaxation time
One can see that only first-order operators occur in (Eq. (12)). FV formulation (Eq. (2)), DG formulation (Eq. (5)), or rDG formulation (Eq. (7)) can be used to integrate the first-order spatial operators.
4. FOHS for Navier-Stokes equations
The FOHS method is straightforward for model equations by introducing the gradients of the unknowns as auxiliary variables. However, for compressible NS equations, the construction of the FOHS becomes trickier. A HNS14 method is first introduced by including viscous stresses and heat fluxed as additional variables [19], but it turns out to obtain reduced order of accuracy in velocity gradients. Later, a HNS 17 method is developed to improve the scheme by including the velocity gradients scaled by viscosity and heat fluxes [20], resulting in the designed order of accuracy in velocity gradients. However, the loss of designed accuracy in density gradients is observed. In order to overcome this issue, an artificial hyperbolic density diffusion term is added to HNS17 to construct HNS20 [12]. The additional term is designed to be small enough such that the scheme can provide expected accurate gradients for density while the continuity equation is not affected. At this point, accurate gradients can be obtained for all primary variables by HNS20 in the pseudo-steady state. However, this efficient construction is not straightforward for other conservative or primary variables. To make the recycling procedure trivial, a new formulation HNS20G [21] was developed recently. Unlike the above-mentioned methods, HNS20G uses the gradients of the primary variables, that is, density, velocity, and temperature, as auxiliary variables to the first-order hyperbolic system. Moreover, with the utilization of the reconstruction method, HNS20G is able to deliver a more accurate solution and gradient while remaining the same degrees of freedom as the conventional DG (P1) method.
4.1 Navier-Stokes equations
The non-dimensionalized Navier–Stokes equations governing unsteady compressible viscous flows can be expressed as
where the summation convention (
where the ratio of the specific heats, is assumed to be constant, that is,
The viscous stress tensor
The Newtonian fluid with the Stokes hypothesis is valid under the current framework. Thus,
where
where
where
According to Fourier’s law, the heat flux vector
where
where
4.2 First-order hyperbolic Navier–Stokes system: HNS20G
The gradients of density, velocity, and temperature are first introduced as auxiliary variables
Then, the governing equations of compressible viscous flows are reformulated as
where
The relaxation times
where the length scale is taken as
Note that for high Reynolds number flows, the length scale should be modified according to the Reynolds number as
where
After writing (Eq. (34)) in the same formulation as (Eq. (12)), either FV formulation (Eq. (2)), DG formulation (Eq. (5)), or rDG formulation (Eq. (7)) can be used to integrate the first-order spatial operators.
4.3 Laminar flow past a sphere
The laminar flow past a sphere is given here to compare the present method with experimental data. The free-stream Reynolds number is taken as
4.4 Laminar flow past a delta wing
A laminar flow at a high angle of attack past a delta wing is considered here. The free-stream Mach number is of
4.5 Von Kármán vortex street behind a circular cylinder
The von Kármán vortex street is one of the most extensively studied cases both experimentally and numerically in fluid dynamics. The initial condition is a uniform-free stream. The free-stream Mach number is taken as
5. Conclusions
High-order reconstructed discontinuous Galerkin methods based on first-order hyperbolic systems for Navier–Stokes equations have been developed and presented. The presented work is based on a new formulation for the hyperbolic Navier–Stokes system, namely HNS20G, which introduces the gradients of density, velocity, and temperature as additional variables. The numerical cases demonstrate that the presented method is compact, efficient, and robust.
Acknowledgments
The authors would like to express their sincere thanks to Prof. Hong Luo at North Carolina State University and Dr. Hiroaki Nishikawa at the National Institute of Aerospace, who has provided many useful insights and thoughts during the development.
References
- 1.
Peshkov I, Romenski E. A hyperbolic model for viscous Newtonian flows. Continuum Mechanics and Thermodynamics. 2016; 28 (1):85-104 - 2.
Dumbser M, Peshkov I, Romenski E, Zanotti O. High order ADER schemes for a unified first order hyperbolic formulation of continuum mechanics: Viscous heat-conducting fluids and elastic solids. Journal of Computational Physics. 2016; 314 :824-862 - 3.
Nishikawa H. First-, second-, and third-order finite-volume schemes for diffusion. Journal of Computational Physics. 2014; 256 :791-805 - 4.
Mazaheri A, Nishikawa H. Improved second-order hyperbolic residual-distribution scheme and its extension to third-order on arbitrary triangular grids. Journal of Computational Physics. 2015; 300 :455-491 - 5.
Lou J, Liu X, Luo H, Nishikawa H. Reconstructed discontinuous Galerkin methods for hyperbolic diffusion equations on unstructured grids. In: 55th AIAA Aerospace Sciences Meeting, AIAA Paper. Vol. 310. Reston, VA: American Institute of Aeronautics & Astronautics (AIAA); 2017 - 6.
Chamarthi AS, Nishikawa H, Komurasaki K. First order hyperbolic approach for anisotropic diffusion equation. Journal of Computational Physics. 2019; 396 :243-263 - 7.
Lou J, Li L, Luo H, Nishikawa H. Reconstructed discontinuous Galerkin methods for linear advection–diffusion equations based on first-order hyperbolic system. Journal of Computational Physics. 2018; 369 :103-124 - 8.
Nishikawa H, Liu Y. Hyperbolic Navier-stokes method for high-Reynolds-number boundary layer flows. In: 55th AIAA Aerospace Sciences Meeting. Reston, VA: American Institute of Aeronautics & Astronautics (AIAA); 2017. p. 0081 - 9.
Anderson JD, Wendt J. Computational Fluid Dynamics. Vol. 206. New York: McGraw-Hill; 1995 - 10.
Toro EF. Riemann Solvers and Numerical Methods for Fluid Dynamics: A Practical Introduction. Berlin/Heidelberg, Germany: Springer-Verlag; 2013 - 11.
Löhner R. Applied Computational Fluid Dynamics Techniques: An Introduction Based on Finite Element Methods. England: John Wiley & Sons; 2008 - 12.
Li L, Lou J, Luo H, Nishikawa H. A new formulation of hyperbolic Navier-stokes solver based on finite volume method on arbitrary grids. In: 2018 Fluid Dynamics Conference. Reston, VA: American Institute of Aeronautics & Astronautics (AIAA); 2018. p. 4160 - 13.
Reddy JN. Introduction to the Finite Element Method. New York City, NY: McGraw-Hill Education; 2019 - 14.
Bassi F, Rebay S. A high-order accurate discontinuous finite element method for the numerical solution of the compressible Navier–stokes equations. Journal of Computational Physics. 1997; 131 (2):267-279 - 15.
Bassi F, Crivellini A, Rebay S, Savini M. Discontinuous Galerkin solution of the Reynolds-averaged Navier–stokes and k–𝜔 turbulence model equations. Computers & Fluids. 2005; 34 (4–5):507-540 - 16.
Kannan R, Wang ZJ. The direct discontinuous Galerkin (DDG) viscous flux scheme for the high order spectral volume method. Computers & Fluids. 2010; 39 (10):2007-2021 - 17.
Cheng J, Yang X, Liu X, Liu T, Luo H. A direct discontinuous Galerkin method for the compressible Navier–stokes equations on arbitrary grids. Journal of Computational Physics. 2016; 327 :484-502 - 18.
Nishikawa H. A first-order system approach for diffusion equation. I: Second-order residual-distribution schemes. Journal of Computational Physics. 2007; 227 (1):315-352 - 19.
Nishikawa H. New-generation hyperbolic Navier-stokes schemes: O (1/h) speed-up and accurate viscous/heat fluxes. In: 20th AIAA Computational Fluid Dynamics Conference. Reston, VA: American Institute of Aeronautics & Astronautics (AIAA); 2011. p. 3043 - 20.
Nishikawa H. Alternative formulations for first-, second-, and third-order hyperbolic Navier-stokes schemes. In: 22nd AIAA Computational Fluid Dynamics Conference. Reston, VA: American Institute of Aeronautics & Astronautics (AIAA); 2015. p. 2451 - 21.
Li L, Lou J, Nishikawa H, Luo H. Reconstructed discontinuous Galerkin methods for compressible flows based on a new hyperbolic Navier-stokes system. Journal of Computational Physics. 2021; 427 :110058 - 22.
Johnson TA. Numerical and Experimental Investigation of Flow Past a Sphere up to a Reynolds Number of 300. Iowa City, Iowa: The University of Iowa; 1996