An unstructured finite volume method based on the projection method combined momentum interpolation with a central scheme for three-dimensional nonhydrostatic turbulent flows
Introduction
Due to the complex physics processes and the limitation of computing power, traditional numerical efforts devoted to the geophysical flow problems, are usually confined into two-dimensional (2D) modeling using the hydrostatic approach [1], [2]. These models can provide rough estimation for future scenarios but lack the capability to handle with three-dimensional (3D) near-field flows such as nonhydrostatic turbulent flows around obstacles in complex geometry. They usually fail in capturing velocity profiles and coherent structure of flows due to the hydrostatic approximation and the absence of correct turbulence modeling [3]. With the development of high performance computer, nowadays LES has been largely used to simulate high turbulent geophysical flows by massive parallel [4], [5], [6].
This paper proposes an accurate, robust and efficient solver for the incompressible Navier–Stokes (N-S) equations using LES in collocated unstructured grids. It is well-known that a collocated grid arrangement for incompressible flows could generate unrealistic pressure oscillations due to the pressure–velocity decoupling, which is known as the checkerboard problem. This problem can be solved using the Momentum Interpolation Method (MIM) [7], [8]. In the past few years, the finite volume method with collocated unstructured grid has been used for both steady and unsteady flows [9], [10], [11], [12]. In these work, the mass flux was calculated by introducing a face-normal velocity, defined at the mid-point of each cell face. This mass flux or the face-normal velocity was interpolated from the cell centers and later corrected by the pressure gradient, which is obtained by the least square method. Depending on the circumstances of different applications, both first order interpolation [13], [14], [15] and second order interpolation [9], [10], [11], [16] could be used.
The originality of this paper is based on the reconstruction of the Projection Methods (PM) using Adams–Bashforth scheme and the Momentum Interpolation Method (MIM) to determine the face-normal velocity combined with central schemes for both convection and diffusion terms. The approximation of the cross diffusion term is improved to handle non-orthogonal, unstructured grids with moderate skewness by introducing an additional correction term. Based on the foregoing method, the second-order accuracy in space and in time is insured in simulating flows at moderate Reynolds numbers.
The structure of the theoretical part is now organized as follows. First, the governing equations including the LES model are presented. The projection method, the time integration, and the finite volume discretization are described in the next section. The numerical results are organized in three sections: accuracy, validation and performance. In the accuracy section, analytical solutions are used to quantify the numerical error of the proposed numerical techniques including the divergence errors, the diffusive and convective terms approximations, and the overall accuracy in time and space. In the validation section, the proposed model is now tested by several well-known benchmarks in comparing the results with experimental solutions, or with other numerical formulations existing in the literature. The test cases include the backward facing step flow, the turbulent channel flow, the lid-driven cavity flow and the flow around a vertical cylinder at a moderate Reynolds number. In the performance section, the convergence of the linear solver is analyzed in terms of iteration numbers and CPU time. Finally, a brief summary and the proposition for future work are also included.
Section snippets
Governing equations
The non-dimensional Navier–Stokes equations for unsteady incompressible viscous flow are given by where the subscripts represent the directions in the Cartesian coordinates, is the non-dimensional velocity component in the direction, is the non-dimensional pressure, and is an external force. The Reynolds number is defined as where the reference scales and are for the velocity and length, respectively.
Projection method and time integration
The projection method (PM) was initially proposed by Chorin [21] to decouple the velocity and pressure fields. To ensure a second order of accuracy in time, the Adams–Bashforth scheme is applied for the diffusion, convection and pressure terms. Thus, Eqs. (9), (10) are explicitly approximated in the following form where superscript donates the variables at time , , and is the time step. In the
Finite-volume method
A second-order unstructured finite-volume method is chosen to discretize Eqs. (14)–(17). The 3D domain is discretized into triangular elements in the horizontal direction and into layers in the vertical direction. As a consequence the domain is discretized into prisms denoted by . Thus, each prism-shaped control volume has five faces (): three with vertical orientation (lateral faces) and two with horizontal orientation (top and bottom face). A schematic plot of a control volume and its
Accuracy of the numerical techniques
The accuracy of the proposed numerical method for incompressible turbulent flows depends a great deal on the approximation of the normal velocity and the outward-normal derivative on the face as well as the type of computational grids. In this section, analytical solutions are used to quantify the numerical error of the proposed numerical techniques. First, the divergence approximation is analyzed in order to study the triangular C-grid divergence noise issue. Next, the numerical performance of
Validation test cases
In this section, the proposed model is tested with several benchmark problems using experimental solutions or numerical simulations from other authors to quantify the numerical error.
Code performance
An efficient solution of the N-S equation is fundamental to the successful development of the proposed three-dimensional nonhydrostatic model for the turbulent flows. In this paper, we employ an explicit formulation based on the Adams–Bashforth scheme. In terms of performance, an implicit time-stepping approach may reduce the simulation time by taking larger values of the time step; however, additional linear system needs to be solved. In a recent publication [26], the authors have proposed a
Conclusions
In this paper, a novel numerical solver has been developed for modeling nonhydrostatic turbulent flows using an unstructured finite-volume method with large eddy simulation. The approximation of values and outward-normal derivative at the cell face in the discretization equation is presented. This approximation will have a direct impact on the accuracy in space of the proposed numerical model. A filtering technique used in the Momentum Interpolation Method avoids the triangular C-grid
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.
Acknowledgments
The authors gratefully acknowledge the INRS-SINAPSE project (ComputeCanada No. 2871), the Mexican Council of Science and Technology project (CONACYT No. 256252), the ANR SSHEAR project (No. 2014-CE03-0011) and the Chinese Scholarship Council (CSC) for their financial supports to do this work. The authors extend special thanks to Électricité de France Recherche & Development (EDF R&D) for their support in providing the access to the computing facility.
References (48)
- et al.
Large-eddy simulation of suspended sediment transport in turbulent channel flow
J. Hydrodyn.
(2013) - et al.
Simulation of droplet dynamic with high density ratio two-phase lattice Boltzmann model on multi-GPU cluster
Comput. & Fluids
(2018) - et al.
Parallel evaluation of quantum algorithms for computational fluid dynamics
Comput. & Fluids
(2018) - et al.
A conservative overlap method for multi-block parallelization of compact finite-volume schemes
Comput. & Fluids
(2017) - et al.
A non-staggered grid, fractional step method for time-dependent incompressible Navier–Stokes equations in curvilinear coordinates
J. Comput. Phys.
(1994) - et al.
A second-order time-accurate finite volume method for unsteady incompressible flow on hybrid unstructured grids
J. Comput. Phys.
(2000) - et al.
A numerical method for large-eddy simulation in complex geometries
J. Comput. Phys.
(2004) - et al.
High-order fluxes for conservative skew-symmetric-like schemes in structured meshes: Application to Compressible Flows
J. Comput. Phys.
(2000) - et al.
An unstructured-grid, finite-volume, nonhydrostatic, parallel coastal ocean simulator
Ocean Model.
(2006) - et al.
A finite volume formulation for transient convection and diffusion equations with unstructured distorted grids and its applications in fluid flow simulations with a collocated variable arrangement
Comput. Methods Appl. Mech. Engrg.
(2013)
Mitigating horizontal divergence checker-board oscillations on unstructured triangular C-grids for nonlinear hydrostatic and nonhydrostatic flows
Ocean Model.
Conservation properties of unstructured staggered mesh schemes
J. Comput. Phys.
A parallel second-order unstructured finite volume method for 3D free-surface flows using a -coordinate
Comput. & Fluids
A superlinearly convergent Mach-uniform finite volume method for the Euler equations on staggered unstructured grids
J. Comput. Phys.
Experimental and numerical investigation of 2D backward-facing step flow
J. Fluids Struct.
High-re solutions for incompressible flow using the Navier–Stokes equations and a multigrid method
J. Comput. Phys.
Turbulent simulation of open channel flow at low Reynolds number
Int. J. Heat Mass Transfer
A novel iterative direct-forcing immersed boundary method and its finite volume applications
J. Comput. Phys.
Quantitative numerical analysis of flow past a circular cylinder at Reynolds number between 50 and 200
J. Fluids Struct.
Depth-averaged two-dimensional numerical modeling of unsteady flow and non-uniform sediment transport in open channels
J. Hydraul. Eng.
Two dimensional river flow and sediment transport model
Environ. Fluid Mech.
Numerical study of the turbulent flow past an airfoil with trailing edge separation
AIAA J.
A quasi-implicit time advancing scheme for unsteady incompressible ?ow. Part I: Validation
Comput. Methods Appl. Mech. Engrg.
A Finite-Volume method for Navier–Stokes equations on unstructured meshes
Numer. Heat Transfer B
Cited by (4)
A conservative level-set/finite-volume method on unstructured grids based on a central interpolation scheme
2021, Journal of Computational PhysicsCitation Excerpt :To mitigate the diffusion effect, second-order schemes must be employed [19]. Moreover, for a triangular C-grid, as presented in this paper, the horizontal divergence should be carefully approximated to prevent high-frequency checkerboard error patterns from affecting the general solution [55,57,59]. Generally, second-order central schemes are designed to deal with problems with enough regularity.
Numerical simulations of the dynamics of axisymmetric compound liquid threads with a phase-field model
2021, European Journal of Mechanics, B/FluidsCitation Excerpt :In this research, the ternary axisymmetric phase-field model is proposed to numerically investigate the dynamics of compound liquid threads, we adopt a practically unconditionally gradient stable method [27] to solve multi-component convective Cahn–Hilliard equation and its discrete form is solved by using multigrid method [28,29]. For Navier–Stokes equation, Chorin’s projection method [30,31] is used. We note that the proposed numerical scheme provides an efficient way to simulate the capillary-driven flow with a 3D axisymmetric characteristic, such as some nanoflows conveying in nanodevices.
Unstructured finite-volume model of sediment scouring due to wave impact on vertical seawalls
2021, Journal of Marine Science and EngineeringThree-Dimensional Hydrostatic Curved Channel Flow Simulations Using Non-Staggered Triangular Grids
2022, Water (Switzerland)