ghost-cell method nfri 16/07/2014
DESCRIPTION
Ghost-Cell method If the shape of a boundary (Plasma/Wall interface) is complex => use of FDM can be delicate Use of Ghost-Cells (GC): points in the solid Value at GC extrapolated from neighbor points (in the Plasma) GC method adapted from Tseng et al. [1]TRANSCRIPT
GHOST-CELL METHODNFRI16/07/2014
Ghost-Cell methodIf the shape of a boundary (Plasma/Wall interface) is com-plex => use of FDM can be delicate• Use of Ghost-Cells (GC): points in the solid • Value at GC extrapolated from neighbor points (in the
Plasma)• GC method adapted from Tseng et al. [1]
BibliographyPaper Dim Reynolds spatial Time Extrapolation Boundary condition otherTseng[1] 2/3 Low/high FVM Linear/quadratic Piecewise linear
segmentNeuman/Dirichlet Mirror
Mittal ARFM [2]
2/3 Low/high FDM Linear/quadratic Rigid/moving Ortho. Projection or middle segment
Majunmbar [3] 2 ~100 - ~1000 FDM implicit Linear/quadratic rigid Mirror
Gibou [4] 3 FDM explicit Linear moving D
Mark [5] 3 ~1 - ~1000 FVM trilinear rigid D/N Mirror
Mittal JCP [6] 3 ~10 - ~1000 FDM Van-Kan linear moving D/N
Berthelsen [7] 2 ~10 - ~100 FDMWENO
RK Cubic / 1D in the same direction
Rigidhighly irregular
D/N
Pan [8] 2/3 ~100 FVM backward Linear rigid
Gao [9] 2 ~100 FDM Adams–Bashforth
Linear or Taylor series
moving D
kim_choi [10] 3 ~100 FDM RK+CN linear moving D/N
Shinn [11] 2/3 ~100 - ~1000 FVM time-marchingfrac-tional-step
linear Rigid/moving D/N mirror
Fluid-Surface Interaction• Solid = inside, fluid = outside
Solid
Fluid
GC
From [1]
• Solid = outside, fluid = inside
(Simple plasma shape assumed)
Tokamak
Plasma(Fluid)
TokamakWall
(Solid)
GC
Tokamak• Boundary = segments• Rigid boundary• 2D (does not depend on Z)• Turbulence: Reynolds >> 1
• Boundary conditions:• ρ =0• ρ.v=0 ?• Bn=0 =>
• Bx = -B// sin α• By = B// cos α
• E = 0.5 * ρ.v2 + P + B// 2/mu0
AlgorithmInitialization step:1- Identify all the GC in the grid2- Compute the interpolation coefficient for each GC
Every time step:3- Use coefficients from 2- to update the GC values
3D = same as 2D (plasma shape does not depend on z)
Algo: Identify GC• Use a mask F(i,j): = 0 for plasma = 1 for GC = 100 for others (no computation)Algo:For all points of the grid (A) cross product with all segments (BC) of the plasma boundary- All z(BAxBC) < 0 => inside plasma => F(A)=0- At least one z(BAxBC) > 0 => outside plasma => F(A)=100For all solid point, test if neighbor = plasma => F(A)=1GC definition depends on WENO method: (md parameter in the code)
Identify Ghost Node - Result• Simple case, with md=4
Algo: Interpolation coefficientGC (G) = extrapolation with 1 projected point (O) and - 2 neighbors (linear) - or 5 neighbors (quadratic)Dirichlet or Neumann=> Need projection of G on boundary and its value(which method? Dirichlet, neumann, linear, quadratic)Ex. Linear dirichlet reconstruction
=>
From [1]
Interpolation coefficient - Result
MirrorProblem if the interpolated point too close of one of the neighbors=> high negative weight => can lead to numerical instability
Use image of ghost nodeInterpolate IThen get G:
From [1]
Algo: update GC values • Use this formulation:
For each GC, at the initial step, need to save: w1, x1,y1,w2, x2,y2, w0, + …,
WENO method• The problem will be solved for each grid point where F(i,j)=0• Near the boundary, the WENO method will use the data of the GC
previously updated• No need to solve the problem on the whole grid
do i=1,nx do i=1,nx do j=1,ny do j=ystart(i),yend(i)
do j=1,ny do j=1,ny do i=1,nx do i=xstart(j),xend(j)
ystart(3)=5 yend(3)=6 xstart(4)=4 xend(4)=5ystart(4)=4 yend(4)=6 xstart(5)=3 xend(5)=6ystart(5)=4 yend(5)=6 xstart(6)=3 xend(6)=6ystart(6)=5 yend(6)=6
No periodicity in X & Y direction (mesh big enough to contain plasma+GC)
Start/end - Result
Bibliography[1]Tseng et al.: A ghost-cell immersed boundary method for flow in complex geometry, Journal of Compu-tational Physics 192 (2003) 593–623[2] R. Mittal and G. Iaccarino. Immersed boundary methods. Annu. Rev. Fluid Mech., 37:239-261, 2005. doi:10.1146 [3] S. Majumdar, G. Iaccarino, and P. Durbin. Rans solvers with adaptive structured boundary noncon-forming grids. Center for Turbulence Research Annual Research Briefs, pages 353-366, 2001.[4] F. Gibou, R.P. Fedkiw, L.T. Cheng, and M. Kang. A Second-Order-Accurate Symmetric Discretization of the Poisson Equation on Irregular Domains. Journal of Computational Physics, 176(1):205-227, 2002. doi:10.1006[5] A. Mark and B.G.M van Wachem. Derivation and validation of a novel implicit second-order accurate immersed boundary method. Journal of Computational Physics, 227(13):6660-6680, 2008. doi:10.1016[6] R. Mittal, H. Dong, M. Bozkurttas, F.M. Najjar, A. Vargas, and A. von Loebbecke. A versatile sharp inter-face immersed boundary method for incompressible flows with complex boundaries. Journal of Compu-tational Physics, 227(10):4825-4852, 2008. doi:10.1016[7] P.A. Berthelsen and O.M. Faltinsen. A local directional ghost cell approach for incompressible viscous flow problems with irregular boundaries. Journal of Computational Physics, 227(9):4354{4397, 2008. doi:10.1016[8] D. Pan and T.T. Shen. Computation of incompressible ows with immersed bodies by a simple ghost cell method. Int. J. Numer. Meth. Fluids, 60(12):1378-1401, 2008. doi:10.1002[9] T. Gao, Y.H. Tseng, and X.Y Lu. An improved hybrid Cartesian/immersed boundary method forfluid-solid flows. Int. J. Numer. Meth. Fluids, 55(12):1189-1211, 2007. doi:10.1002[10] D. Kim and H. Choi. Immersed boundary method for flow around an arbitrarily moving body. Journal of Computational Physics, 212(2):662-680, 2006. doi:10.1016[11] A.F. Shinn, M.A. Goodwin, and S.P. Vanka. Immersed boundary computations of shear- and buoyan-cydriven flows in complex enclosures. International Journal of Heat and Mass Transfer, 52(17-18):4082-4089, 2009. doi:10.1016