Development of discrete gas kinetic scheme for simulation of 3D viscous incompressible and compressible flows
Introduction
In the computational fluid dynamics (CFD) community, the gas kinetic scheme (GKS) [1], [2], [3], [4], [5], [6], [7], [8] has been developed into a powerful tool for simulation of fluid flows. Like the popular Riemann solvers [9], [10], [11], [12], [13], [14], the GKS is intended to evaluate the numerical fluxes at the cell interface. As a result, the GKS can be easily embedded in the existing Riemann solver-based CFD codes. On the other hand, different from the conventional Riemann solvers, the GKS treats the inviscid and viscous fluxes as a single entity instead of using a Riemann solver to evaluate the inviscid fluxes and a smooth function approximation to compute the viscous fluxes. In addition, the GKS has a clear physical process (streaming and collision) for evaluation of conservative variables and numerical fluxes, and the scheme can be well applied to both incompressible and compressible flows without unphysical phenomenon such as “Carbuncle phenomenon”. This makes the GKS be more appealing than conventional Riemann solvers.
In the GKS, the local solution of continuous Boltzmann equation with the Maxwellian distribution function is usually used to calculate the numerical fluxes at the cell interface [15], [16], [17], [18], [19], [20], [21], [22]. Due to complexity of the Maxwellian distribution function, these GKSs are usually more complicated and less efficient than the conventional Riemann solvers [9], [10], [11], [12], [13], [14]. In particular, for the 3D viscous flows, the derivation and implementation of the GKS can be very tedious [20], [21]. To improve the computational efficiency of the GKS and simplify its derivation, Shu and his coworkers [23] proposed the sphere function-based GKS for simulation of 3D inviscid compressible flows. In the method, the Maxwellian distribution function is first simplified into the sphere function with assumption that all the particles are concentrated on a spherical surface. Thus, the integrals in the infinity domain of phase velocity space for the Maxwellian function can be reduced to the surface integrals along the spherical surface for the sphere function. As shown in [23], since the expressions of the sphere function-based GKS are relatively simple, it is easier to be implemented than the corresponding Maxwellian distribution function-based GKS. However, the presented sphere function-based GKS can only be used to simulate inviscid compressible flows in the previous work. In theory, we can follow the idea of developing the circular function-based GKS for viscous flows [8] to extend the sphere function-based GKS to viscous flow regime. However, the resultant formulations could be very complicated, which may not be easy for the application by new users. To develop a simple and efficient sphere function-based GKS for simulation of 3D viscous flows motivates the present work.
For the sphere function-based GKS [23], the integrals for conservation forms of moments are along the spherical surface. In addition, the Maxwellian distribution function is reduced to a simple form, which only depends on macroscopic flow variables on the spherical surface. These simplifications enable us to represent the spherical surface in the phase velocity space by using certain discrete points. That is, the integrals along the spherical surface for conservation forms of moments can be approximated by integral quadrature. In this process, the basic requirement is that the conservation forms of moments for the sphere function-based GKS, which are needed to recover 3D Navier–Stokes equations, can be exactly satisfied by weighted summation of distribution functions at discrete points. It was found in this work that, the model with 8 discrete points on the spherical surface, which forms the D3Q8 discrete velocity model, can exactly match the integrals. In this way, the macroscopic flow variables and numerical fluxes can be computed by weighted summations of distribution functions at 8 discrete points, and the application of complicated formulations resultant from integrals is avoided.
The rest of the paper is organized as follows. The development of D3Q8 model from sphere function-based GKS is presented in Section 2. In Section 3, the developed D3Q8 model is utilized to solve the 3D Navier–Stokes equations. Numerical examples and discussions are arranged in Section 4. In the last section (Section 5), some concluding remarks are given.
Section snippets
Sphere function-based GKS and 7 conservation forms of moments
The existing sphere function-based GKS [23] is developed to simulate the 3D inviscid compressible flows. Thus, only 5 conservation forms of moments for recovering the Euler equations are used. To develop discrete velocity model for simulation of viscous flows, the 7 conservation forms of moments for recovering the Navier–Stokes equations are needed. So, at beginning, the sphere function-based GKS and its 7 conservation forms of moments for recovering the Navier–Stokes equations are introduced.
3D Navier–Stokes equations discretized by finite volume method
For the convenience of derivation, we first introduce the local coordinate system. In the local coordinate system, direction 1 is taken as the normal direction pointing always outwards of the cell interface, directions 2 and 3 are chosen as two tangential directions of the cell interface, which are mutually orthogonal. The discrete form of Navier–Stokes equations given by finite volume method (FVM) can be written as [23] where I is the index of a control volume, and
Numerical examples
In this section, the developed discrete GKS is validated by simulating some 3D incompressible and compressible viscous flows. For temporal discretization of Eq. (10), the three-stages Runge–Kutta method [13] is used. In order to get a stable numerical solution for compressible flows, some numerical dissipation is needed to be incorporated into the solver. A simple way to incorporate numerical dissipation into the present scheme is to modify the dimensionless collision time given by Eq. (16) as
Conclusions
In this work, the sphere function-based GKS proposed in our previous work for simulation of inviscid compressible flows is extended to solve the viscous incompressible and compressible flows. Different from the existing sphere function-based GKS [23], which directly uses the continuous sphere function to evaluate numerical fluxes at the cell interface, the present scheme is based on the discrete velocity model developed by the sphere function. By using certain discrete points to represent the
References (37)
Gas-kinetic schemes for compressible flow of real gases
Comput. Math. Appl.
(2001)- et al.
A multidimensional gas-kinetic BGK scheme for hypersonic viscous flow
J. Comput. Phys.
(2005) Progress in gas-kinetic upwind schemes for the solution of Euler/Navier–Stokes equations – I: overview
Comput. Fluids
(2012)- et al.
WENO-enhanced gas-kinetic scheme for direct simulations of compressible transition and turbulence
J. Comput. Phys.
(2013) - et al.
Circular function-based gas-kinetic scheme for simulation of inviscid compressible flows
J. Comput. Phys.
(2013) - et al.
A simple distribution function-based gas-kinetic scheme for simulation of viscous incompressible and compressible flows
J. Comput. Phys.
(2014) Towards the ultimate conservative difference scheme V. A second order sequel to Godunov's method
J. Comput. Phys.
(1979)Approximate Riemann solvers, parameter vectors, and difference schemes
J. Comput. Phys.
(1981)- et al.
A new flux splitting scheme
J. Comput. Phys.
(1993) Direct simulation methods for compressible inviscid ideal-gas flow
J. Comput. Phys.
(1980)
Kinetic flux vector splitting for Euler equations
Comput. Fluids
Kinetic flux–vector splitting for the Navier–Stokes equations
J. Comput. Phys.
Development of an improved gas-kinetic BGK scheme for inviscid and viscous flows
J. Comput. Phys.
A gas-kinetic BGK scheme for the Navier–Stokes equations and its connection with artificial dissipation and Godunov method
J. Comput. Phys.
A three-dimensional multidimensional gas-kinetic scheme for the Navier–Stokes equations under gravitational fields
J. Comput. Phys.
Implicit gas-kinetic BGK scheme with multigrid for 3D stationary transonic high-Reynolds number flows
Comput. Fluids
Explicit formulations of gas-kinetic flux solver for simulation of incompressible and compressible viscous flows
J. Comput. Phys.
A three-dimensional explicit sphere function-based gas-kinetic flux solver for simulation of inviscid compressible flows
J. Comput. Phys.
Cited by (29)
Hydrodynamic characterization of manta rays in bowed gliding state
2024, Ocean EngineeringThree-dimensional high-order finite-volume method based on compact WENO reconstruction with hybrid unstructured grids
2023, Journal of Computational PhysicsA novel solver for simulation of flows from continuum regime to rarefied regime at moderate Knudsen number
2020, Journal of Computational PhysicsCitation Excerpt :To simplify the moment method and extend its application to complex problems, some improvement is needed. Recently, the gas kinetic flux solver (GKFS) presented by Shu and his coworkers [39–44] for simulation of flows in the continuum flow regime gives us a great inspiration to retain the advantage of the moment method and greatly simplify the solution process. In GKFS, the finite volume method (FVM) is applied to discretize the macroscopic governing equations and the numerical fluxes at the cell interface are obtained by the local solution of Boltzmann equation.
An implicit simplified sphere function-based gas kinetic scheme for simulation of 3D incompressible isothermal flows
2018, Computers and FluidsCitation Excerpt :This work presents an implicit simplified SGKS for simulation of 3D incompressible isothermal flows. It is a further improvement to our previous works [22,23], in which the SGKS is developed mainly for simulation of compressible flows. For compressible flows, the integral domain along the spherical surface for deriving the explicit formulations is usually not symmetric at the cell interface, while for incompressible flows, the sphere at the cell interface can be approximately considered to be symmetric due to incompressible limit.