LDG2: A Variant of the LDG Flux Formulation for the Spectral Volume Method

The local discontinuous Galerkin (LDG) viscous flux formulation was originally developed by Cockburn and Shu for the discontinuous Galerkin setting and later extended to the spectral volume setting by Wang and his collaborators. Unlike the penalty formulations like the interior penalty and the BR2 schemes, the LDG formulation requires no length based penalizing terms and is compact. However, computational results using LDG are dependant of the orientation of the faces especially for unstructured and non uniform grids. This results in lower solution accuracy and stiffer stability constraints as shown by Kannan and Wang. In this paper, we develop a variant of the LDG, which not only retains its attractive features, but also vastly reduces its unsymmetrical nature. This variant (aptly named LDG2), displayed higher accuracy than the LDG approach and has a milder stability constraint than the original LDG formulation. In general, the 1D and the 2D numerical results are very promising and indicate that the approach has a great potential for 3D flow problems.     

A Semi-Lagrangian Method for Turbulence Simulations Using Mixed Spectral Discretizations

We present a semi-Lagrangian method for integrating the three-dimensional incompressible Navier–Stokes equations. We develop stable schemes of second-order accuracy in time and spectral accuracy in space. Specifically, we employ a spectral element (Jacobi) expansion in one direction and Fourier collocation in the other two directions. We demonstrate exponential convergence for this method, and investigate the non-monotonic behavior of the temporal error for an exact three-dimensional solution. We also present direct numerical simulations of a turbulent channel-flow, and demonstrate the stability of this approach even for marginal resolution unlike its Eulerian counterpart.     

On the application of the minimum polynomial extrapolation method to incompressible flows with heat transfer

In this paper, the Minimum Polynomial Extrapolation method (MPE) is used to accelerate the convergence of the Characteristic–Based–Split (CBS) scheme for the numerical solution of steady state incompressible flows with heat transfer. The CBS scheme is a fractional step method for the solution of the Navier–Stokes equations while the MPE method is a vector extrapolation method which transforms the original sequence into another sequence converging to the same limit faster then the original one without the explicit knowledge of the sequence generator. The developed algorithm is tested on a two-dimensional benchmark problem (buoyancy–driven convection problem) where the Navier–Stokes equations are coupled with the temperature equation. The obtained results show the feature of the extrapolation procedure to the CBS scheme and the reduction of the computational time of the simulation.     

A Parallel Overset Grid High-Order Flow Solver for Large Eddy Simulation

This work describes the development and validation of a parallel high-order compact finite difference Navier–Stokes solver for application to large-eddy simulation (LES) and direct numerical simulation. The implicit solver can employ up to sixth-order spatial formulations and tenth-order filtering. The parallelization of the solver is founded on the overset grid technique. LES were then performed for turbulent channel flow with Reynolds numbers ranging from Re _τ=180 to 590, and flow past a circular cylinder with a transitional wake at Re _D =3900. The channel flow solutions were obtained using both an implicit LES (ILES) approach and a dynamic sub-grid scale model. The ILES method obtained virtually identical solutions at half the computational cost. The original vector and new parallel solvers produce indistinguishable mean flow solutions for the circular cylinder. Repeating the cylinder simulation on a much finer mesh resulted in significantly better agreement with experimental data in the near wake than the coarse grid solution and other previous numerical studies.     

Fourth-Order Runge–Kutta Schemes for Fluid Mechanics Applications

Multiple high-order time-integration schemes are used to solve stiff test problems related to the Navier–Stokes (NS) equations. The primary objective is to determine whether high-order schemes can displace currently used second-order schemes on stiff NS and Reynolds averaged NS (RANS) problems, for a meaningful portion of the work-precision spectrum. Implicit–Explicit (IMEX) schemes are used on separable problems that naturally partition into stiff and nonstiff components. Non-separable problems are solved with fully implicit schemes, oftentimes the implicit portion of an IMEX scheme. The convection–diffusion-reaction (CDR) equations allow a term by term stiff/nonstiff partition that is often well suited for IMEX methods. Major variables in CDR converge at near design-order rates with all formulations, including the fourth-order IMEX additive Runge–Kutta (ARK₂) schemes that are susceptible to order reduction. The semi-implicit backward differentiation formulae and IMEX ARK₂ schemes are of comparable efficiency. Laminar and turbulent aerodynamic applications require fully implicit schemes, as they are not profitably partitioned. All schemes achieve design-order convergence rates on the laminar problem. The fourth-order explicit singly diagonally implicit Runge–Kutta (ESDIRK4) scheme is more efficient than the popular second-order backward differentiation formulae (BDF2) method. The BDF2 and fourth-order modified extended backward differentiation formulae (MEBDF4) schemes are of comparable efficiency on the turbulent problem. High precision requirements slightly favor the MEBDF4 scheme (greater than three significant digits). Significant order reduction plagues the ESDIRK4 scheme in the turbulent case. The magnitude of the order reduction varies with Reynolds number. Poor performance of the high-order methods can partially be attributed to poor solver performance. Huge time steps allowed by high-order formulations challenge the capabilities of algebraic solver technology.     

A two-port model for wave propagation along a long circular microchannel

A compact model for oscillatory flow in a long microchannel with a circular cross-section is derived from the linearised Navier–Stokes equations. The resulting two-port model includes the effects of viscosity due to rarefied gas in the slip flow regime, inertia, compressibility and losses due to heat exchange. Both an acoustic impedance T network and an acoustic admittance Π network are presented for implementation in system level and circuit simulation tools. Also, reduced T and Π networks with constant component values are given to be used in the low frequency region. They are useful in time domain simulations, too. To verify the analytical model, simulations with a harmonic finite element solver for acoustic viscous flow are performed for microchannels exploiting the axisymmetry. The simulation results with both open and closed outlet conditions are compared with the two-port model with excellent agreement. Contribution of the slip conditions and the accuracy of the simple model are demonstrated.     

Adherence and bouncing of liquid droplets impacting on dry surfaces

The paper explores liquid drop dynamics over a solid surface, focusing on adherence and bouncing phenomena. The study relies on detailed interface tracking simulations using the Level Set approach incorporated within a Navier–Stokes solver. The investigation deals with moderate Reynolds number droplet flows, for which two-dimensional axisymmetric simulations can be performed. The modelling approach has been validated against experiments for axisymmetric and full three-dimensional impact upon dry surfaces. A drop-impact regime map is generated for axisymmetric conditions, in which the impact dynamics is characterized as a function of Weber number and equilibrium contact angle, based on about 60 simulations. The detailed simulations also helped validate a new mechanistic model based on energy-balance analysis, delimiting the boundary between adherence and bouncing zones at low Weber numbers. The mechanistic model is only valid for moderate droplet Reynolds numbers and it complements existing models for higher Reynolds numbers.     

The direct discontinuous Galerkin (DDG) viscous flux scheme for the high order spectral volume method

The direct discontinuous Galerkin (DDG) method was developed by Liu and Yan to discretize the diffusion flux. It was implemented for the discontinuous Galerkin (DG) formulation. In this paper, we perform four tasks: (i) implement the direct discontinuous Galerkin (DDG) scheme for the spectral volume method (SV) method, (ii) design and implement two variants of DDG (called DDG2 and DDG3) for the SV method, (iii) perform a Fourier type analysis on both methods when solving the 1D diffusion equation and combine the above with a non-linear global optimizer, to obtain modified constants that give significantly smaller errors (in 1D), (iv) use the above coefficients as starting points in 2D. The dissipation properties of the above schemes were then compared with existing flux formulations (local discontinuous Galerkin, Penalty and BR2). The DDG, DDG2 and DDG3 formulations were found to be much more accurate than the above three existing flux formulations. The accuracy of the DDG scheme is heavily dependent on the penalizing coefficient for the odd ordered schemes. Hence a loss of accuracy was observed even for mildly non-uniform grids for odd ordered schemes. On the other hand, the DDG2 and DDG3 schemes were mildly dependent on the penalizing coefficient for both odd and even orders and retain their accuracy even on highly irregular grids. Temporal analysis was also performed and this yielded some interesting results. The DDG and its variants were implemented in 2D (on triangular meshes) for Navier–Stokes equations. Even the non-optimized versions of the DDG displayed lower errors than the existing schemes (in 2D). In general, the DDG and its variants show promising properties and it indicates that these approaches have a great potential for higher dimension flow problems.     

Numerical Study of a Modified Time-Stepping θ-Scheme for Incompressible Flow Simulations

In [Turek (1996). Int. J. Numer. Meth. Fluids 22, 987–1011], we had performed numerical comparisons for different time stepping schemes for the incompressible Navier–Stokes equations. In this paper, we present the numerical analysis in the context of the Navier–Stokes equations for a modified time-stepping θ-scheme which has been recently proposed by Glowinski [Glowinski (2003). In: Ciarlet, P. G., and Lions, J. L. (eds.), Handbook of Numerical Analysis, Vol. IX, North-Holland, Amsterdam, pp. 3–1176]. Like the well-known classical Fractional-Step-θ-scheme which had been introduced by Glowinski [Glowinski (1985). In Murman, E. M. and Abarbanel, S. S. (eds.), Progress and Supercomputing in Computational Fluid Dynamics, Birkh?user, Boston MA; Bristeau et al. (1987). Comput. Phys. Rep. 6, 73–187], too, and which is still one of the most popular time stepping schemes, with or without operator splitting techniques, this new scheme consists of 3 substeps with nonequidistant substepping to build one macro time step. However, in contrast to the Fractional-Step-θ-scheme, the second substep can be formulated as an extrapolation step for previously computed data only, and the two remaining substeps look like a Backward Euler step so that no expensive operator evaluations for the right hand side vector with older solutions, as for instance in the Crank–Nicolson scheme, have to be performed. This modified scheme is implicit, strongly A-stable and second order accurate, too, which promises some advantageous behavior, particularly in implicit CFD simulations for the nonstationary Navier–Stokes equations. Representative numerical results, based on the software package FEATFLOW [Turek (2000). FEATFLOW Finite element software for the incompressible Navier–Stokes equations: User Manual, Release 1.2, University of Dortmund] are obtained for typical flow problems with benchmark character which provide a fair rating of the solution schemes, particularly in long time simulations.Dedicated to David Gottlieb on the occasion of his 60th anniversary     

Study of subsonic�Csupersonic gas flow through micro/nanoscale nozzles using unstructured DSMC solver

We use an extended direct simulation Monte Carlo (DSMC) method, applicable to unstructured meshes, to numerically simulate a wide range of rarefaction regimes from subsonic to supersonic flows through micro/nanoscale converging–diverging nozzles. Our unstructured DSMC method considers a uniform distribution of particles, employs proper subcell geometry, and follows an appropriate particle tracking algorithm. Using the unstructured DSMC, we study the effects of back pressure, gas/surface interactions (diffuse/specular reflections), and Knudsen number on the flow field in micro/nanoscale nozzles. If we apply the back pressure at the nozzle outlet, a boundary layer separation occurs before the outlet and a region with reverse flow appears inside the boundary layer. Meanwhile, the core region of inviscid flow experiences multiple shock-expansion waves. In order to accurately simulate the outflow, we extend a buffer zone at the nozzle outlet. We show that a high viscous force creation in the wall boundary layer prevents any supersonic flow formation in the divergent part of the nozzle if the Knudsen number exceeds a moderate magnitude. We also show that the wall boundary layer prevents forming any normal shock in the divergent part. In reality, Mach cores would appear at the nozzle center followed by bow shocks and expansion region. We compare the current DSMC results with the solution of the Navier–Stokes equations subject to the velocity slip and temperature jump boundary conditions. We use OpenFOAM as a compressible flow solver to treat the Navier–Stokes equations.     

The comparison of incomplete sensitivities and Genetic algorithms applications in 3D radial turbomachinery blade optimization

In the present work, a centrifugal pump impeller’s blades shape was redesigned to reach a higher efficiency in turbine mode using two different optimization algorithms: one is a local method as incomplete sensitivities–gradient based optimization algorithm coupled by 3D Navier–Stokes flow solver, and another is a global method as Genetic algorithms and artificial neural network coupled by 3D Navier–Stokes flow solver. New impeller was manufactured and tested in the test rig. Comparison of the local optimization method results with the global optimization method results showed that the gradient based method has detected the global optimum point. Experimental results confirmed the numerical efficiency improvement in all measured points. This study illustrated that the developed gradient based optimization method is efficient for 3D radial turbomachinery blade optimization.     

Refinement of flexible space–time finite element meshes and discontinuous Galerkin methods

In this paper we present an algorithm to refine space–time finite element meshes as needed for the numerical solution of parabolic initial boundary value problems. The approach is based on a decomposition of the space–time cylinder into finite elements, which also allows a rather general and flexible discretization in time. This also includes adaptive finite element meshes which move in time. For the handling of three-dimensional spatial domains, and therefore of a four-dimensional space–time cylinder, we describe a refinement strategy to decompose pentatopes into smaller ones. For the discretization of the initial boundary value problem we use an interior penalty Galerkin approach in space, and an upwind technique in time. A numerical example for the transient heat equation confirms the order of convergence as expected from the theory. First numerical results for the transient Navier–Stokes equations and for an adaptive mesh moving in time underline the applicability and flexibility of the presented approach.     

On the viscous–viscous and the viscous–inviscid interactions in Computational Fluid Dynamics

We consider here the exterior boundary value problem for compressible viscous flow around airfoils. In a first approximation, the viscosity effects are neglected at some distance to the airfoil. The unbounded domain is decomposed by an artificial boundary into a bounded computational domain (near field) and an associated far field. The complete system of conservation laws, modelling viscous flow in the near field is coupled with simplified models for inviscid flow in the far field. The use of the heterogeneous domain decomposition method including physically and mathematically justified transmission conditions at the artificial interface provides one with a quite accurate approximate solution, modelling the viscous–inviscid interaction between the two model zones. However, such a solution does not take into account the viscosity in the far field and does not satisfy the natural transmission conditions at the artificial interface (i.e. continuity of the solution and of the normal flux). In order to get some information for the a-posteriori improvement of this solution, we introduce one-dimensional transmission-boundary value problems, obtained by an appropriate dimensional reduction of the coupled problems from CFD. The one-dimensional problems are analyzed in the framework of singular perturbation theory. We consider formal asymptotic expansions to construct appropriate boundary layer corrections of the coupled problem modelling the viscous–inviscid interaction. Our one-dimensional analysis seems to allow an extension to higher dimensions and therefore could be used in the computation of the solution to the compressible Navier–Stokes problem by updating the solution of the approximation by a (degenerate) Navier–Stokes/Euler problem with boundary layer viscosity correction terms. Received: 22 February 1999 / Accepted: 17 June 1999     

An implicit matrix-free Discontinuous Galerkin solver for viscous and turbulent aerodynamic simulations

This paper presents some recent advancements of the computational efficiency of a Discontinuous Galerkin (DG) solver for the Navier–Stokes (NS) and Reynolds Averaged Navier Stokes (RANS) equations. The implementation and the performance of a Newton–Krylov matrix-free (MF) method is presented and compared with the matrix based (MB) counterpart. Moreover two solution strategies, developed in order to increase the solver efficiency, are discussed and experimented. Numerical results of some test cases proposed within the EU ADIGMA (Adaptive Higher-Order Variational Methods for Aerodynamic Applications in Industry) project demonstrate the capabilities of the method.     

A finite element based level set method for two-phase incompressible flows

We present a method that has been developed for the efficient numerical simulation of two-phase incompressible flows. For capturing the interface between the phases the level set technique is applied. The continuous model consists of the incompressible Navier–Stokes equations coupled with an advection equation for the level set function. The effect of surface tension is modeled by a localized force term at the interface (so-called continuum surface force approach). For spatial discretization of velocity, pressure and the level set function conforming finite elements on a hierarchy of nested tetrahedral grids are used. In the finite element setting we can apply a special technique to the localized force term, which is based on a partial integration rule for the Laplace–Beltrami operator. Due to this approach the second order derivatives coming from the curvature can be eliminated. For the time discretization we apply a variant of the fractional step θ-scheme. The discrete saddle point problems that occur in each time step are solved using an inexact Uzawa method combined with multigrid techniques. For reparametrization of the level set function a new variant of the fast marching method is introduced. A special feature of the solver is that it combines the level set method with finite element discretization, Laplace–Beltrami partial integration, multilevel local refinement and multigrid solution techniques. All these components of the solver are described. Results of numerical experiments are presented.     

Lattice Boltzmann based PDE solver on the GPU

In this paper, we propose a hardware-accelerated PDE (partial differential equation) solver based on the lattice Boltzmann model (LBM). The LBM is initially designed to solve fluid dynamics by constructing simplified microscopic kinetic models. As an explicit numerical scheme with only local operations, it has the advantage of being easy to implement and especially suitable for graphics hardware (GPU) acceleration. Beyond the Navier–Stokes equation of fluid mechanics, a typical LBM can be modified to solve the parabolic diffusion equation, which is further used to solve the elliptic Laplace and Poisson equations with a diffusion process. These PDEs are widely used in modeling and manipulating images, surfaces and volumetric data sets. Therefore, the LBM scheme can be used as an GPU-based numerical solver to provide a fast and convenient alternative to traditional implicit iterative solvers. We apply this method to several examples in volume smoothing, surface fairing and image editing, achieving outstanding performance on contemporary graphics hardware. It has the great potential to be used as a general GPU computing framework for efficiently solving PDEs in image processing, computer graphics and visualization.     

Real time simulation of a tornado

We present a novel method for simulating a tornado scene and its damage on the environment in real time, which is recognized as a challenging task for researchers of computer graphics. The method adopts a Reynold-average two-fluid model (RATFM) for modeling the motion of a tornado. In RATFM, the air flow (wind field) is simulated by Reynold-average Navier–Stokes equations. The motion of dust particles is approximated as a continuous fluid and is modeled by non-viscosity Navier–Stokes equations. An interaction force is introduced to simulate the interaction between these two-fluid systems efficiently. Considering the data structure of our method, we design a RATFM solver on the GPU to achieve real time simulation. We also adopt new features of the GPU to accelerate our algorithm. Then, an efficient method is proposed to simulate the tornado’s interaction with surrounding large objects such as a car, a bus, a house, etc. In our model, the objects in the tornado scene are represented by connected voxels and a corresponding graph storing the link information. Compared with the photographs of real tornado displays, our simulated results are quite satisfactory.     

Least-Squares Spectral Collocation for the Navier–Stokes Equations

A least-squares spectral collocation formulation for the Navier–Stokes problem is presented. By this new approach the well known Babuska–Brezzi condition can be avoided. Here we are able to employ polynomials of the same degree both for the velocity components and for the pressure. The collocation conditions and the boundary conditions lead to a overdetermined system which can be efficiently solved by least-squares. The solution technique will only involve symmetric positive definite linear systems. The numerical simulations confirm the usual exponential rate of convergence for the spectral scheme.     

A <Emphasis Type="Italic">p</Emphasis>-Adaptive Local Discontinuous Galerkin Level Set Method for Willmore Flow

The level set method is often used to capture interface behavior in two or three dimensions. In this paper, we present a combination of a local discontinuous Galerkin (LDG) method and a level set method for simulating Willmore flow. The LDG scheme is energy stable and mass conservative, which are good properties compared with other numerical methods. In addition, to enhance the efficiency of the proposed LDG scheme and level set method, we employ a p-adaptive local discontinuous Galerkin technique, which applies high order polynomial approximations around the zero level set and low order ones away from the zero level set. A major advantage of the level set method is that the topological changes are well defined and easily performed. In particular, given the stiffness and high nonlinearity of Willmore flow, a high order semi-implicit Runge–Kutta method is employed for time discretization, which allows larger time steps. These equations at the implicit time level are linear, we demonstrate an efficient and practical multigrid solver to solve the equations. Numerical examples are given to illustrate that the combination of the LDG scheme and level set method provides an efficient and practical approach to simulate the Willmore flow.