a mesh-less method for solutions of the stefan problem
DESCRIPTION
A Mesh-less method for solutions of the Stefan Problem. Vaughan Voller. If A man knows he is going to be hanged tomorrow it concentrates the mind wonderfully— Dr Samuel Johnson 1709-1784. Preliminary Results. Essence of a Numerical Solution. Cover Domain with - PowerPoint PPT PresentationTRANSCRIPT
A Mesh-less method for solutions of the Stefan Problem
Preliminary Results
Vaughan Voller
If A man knows he is going to be hanged tomorrow it concentrates the mind wonderfully—
Dr Samuel Johnson 1709-1784
0T2
Essence of a Numerical Solution
1. Cover Domain with Field of NODES with locations xi
2. Unknowns are nodal values of dependent variable
iT)(T ix
3. By appropriate APPROXIMATIONSof Governing Equation Obtain a SETOf Discrete Algebraic equationsThat relate the nodal value at P to Values at the neighbors
P
nb
nbnbPP TaTa
Process is facilitated by
DATA Structure of node points
Approximation Process is facilitated by-- DATA Structure of node points
P
The most Straight Forward Is a STRUCTURED GRID OF NODES
Location of anynode point given aRow, Column index
If Grid is “square” immediate application of Taylor Series
0T4TTTTT ijj1ij1i1ij1ij2 Row i
Col. j
Approximation Process is facilitated by-- DATA Structure of node points
A more flexible approach is use Nodes to DefineAn UNSTRUCTURED MESH OF ELEMENTS
In Each Element obtain a Continuous Approximation
1
2
3
332211e T)y,x(NT)y,x(NT)y,x(N)y,x(T
Use this approximation with governing equatione.g. in CVFEM--use
0dAT n
n
Approximation Process is facilitated by-- DATA Structure of node points
The most flexible approach is to have no mesh
Very limited restriction on placement of nodes
But may have to be inventive
In arriving at sound discrete equations
nb
nbnbPP TaTa
---CLOUDS of Near Neighbor Nodes
Increasing Flexibility
GRID---Structure
Easy ApproximationEfficient solution Ax=b —
Difficult to Fit Geometry
MESH---FEM
More Complex approximationLess Efficient solution Ax=b —
Easy to Fit GeometryCould be Difficult to adapt
“CLOUD”---SPH
“Poor” approximationEven Less Efficient solution Ax=b —
Very Easy to Fit GeometryEasy to adapt
Simplistic Summary
Corrective Smooth Particle Method CSPM -related to SPHChen et al IJNME 1999
P
..)yy(yT)xx(
xTT)(T PPP
x
Taylor Series about node P
Multiply by Weighting Factor associated with node P
..)yy(yT)(W)xx(
xT)(WT)(W)(T)(W PPPPPPP
xxxxx
..)yy(yT)(W)xx(
xT)(WT)(W)(T)(W PPPPPPP
xxxxx
Properties of W
P2hp
nbP Rminh 0.5
Symmetric about point P
Finite region of support
Differentiable
2r,0
2r1,r2h41
1r,rr1h1
)r(W 32
3432
23
2
r=R/h
Char. Length multiple of nearest neig. distance
Corrective Smooth Particle Method CSPM -related to SPHChen et al IJNME 1999
P
..)yy(yT)xx(
xTT)(T PPP
x
Taylor Series about node P
Multiply by Weighting Factor associated with node P
..)yy(yT)(W)xx(
xT)(WT)(W)(T)(W PPPPPPP
xxxxx
Integrate over support ..)yy(
yT)(W)xx(
xT)(WT)(WdV)(T)(W PPPPPPP
xxxxx
dVW
dV)(T)(WT
P
P
P
xxsimilar manipulations for first derivatives
dVW
dV)(T)(WT
P
P
P
xx
nbP Rminh
P
If Rnb is the radial distance to the neighbors of P
0.5
Corrective Smooth Particle Method CSPM -related to SPHChen et al IJNME 1999 Integrate numerically
Using particles as integration points
nodeskkP
nodeskkkP
P Vol)(W
VolT)(WT
k
k
x
x
2hp
2r,0
2r1,r2h41
1r,rr1h1
)r(W 32
3432
23
2
r=R/h
nbnb
nbnbnbP
1w
,TwT
Critical Feature
Weak point --- 2kk hVol
SolidTi < Tm
a.) time t = 0
Application to The Stefan Problem--- General problem of Interest
SolidT < Tm
b.) time t > 0
LiquidT > Tm
liquid-solid interface T = Tm
T=Ta >Tm
n
At time t>=0 apply a fixed temperatureT=Ta >Tm to a patch of the boundary so as to cause a melt region that grows with time
Objective track the movement of the melt front melt
Initial state Insulated region containing solid atTemperature T; < Tm (melt temp)
l
2
,
Assume constant density and specific heat c — but jump in conductivity K.
with time scale and space scales
Governing Equations
ma
mD
TTTTT
L1
H)TT(cS ma
t
Dimensionless temperature
And dimensionless grouping , H-latent heat, St-Stefan number, L dim. Lat. heat
T=0g=0
g=1
TtT 2
TKtT 2
nvn tS
1TKT
liquid-solid interface
T = 0
n
Two-Domain Stefan Model Diffusive Interface—Single Domain
)TK̂(tH
Phase change occurs smoothly acrossA “narrow” temperature range
gLTH
K)g1(gK̂
Particular Versions
s(t)
T=1 T= 0.5
K=0.25
L=1
one-dimension
Has analytical solution
T=1
Ti = 0
Melting of unit cylinder, initially at phaseChange temperature.
)T(tH
Solve in Cartesian
Check with fine grid FD solutionUsing radial symmetry
)xTK(
xtH
Two Dimension
CSPM Solution
SCPMPTq
SCPMPP
newP qtHH
q)T(tH
Use CSPM approximation of derivative twiceBackward Euler (explicit) in time
PP
nbnbnbP
newP TaTatHH
Can be manipulated into general form
Data Structure
Global Number nodes 1-----n
Identify “cloud” of neighborsassociated with each node----
make a list of nodes (global numbers)that fall within a radial distance 2hP
2r,0
2r1,r2h41
1r,rr1h1
)r(W 32
3432
23
2
P2hp
LHLH
00HH
TnewP
newP
newP
newP
newP
T=1 T=-0.5
K = 0.25
t = 0.002h = 0.033-0.05
One-D solution
Front Movement
o o o CSPM
Analytical
Temperature HistoriesAt ref points
Plateau at phase change temp.A feature in all fixed grid solutions
Temperature Profileat time t =2.8
T=1
Ti = 0
)T(tH
For the 2-D problem Need to consider a means of placing out points
Two Methods
Discretization: Two steps
Put points along boundary of domain—with equal arc spacing r Make a structured mesh with spacing r
Then Lay boundary Mesh Over Structure MeshAdd structured points to SPH node List if they are a distance INSIDE boundary
SPH Nodes are a List i=1 to nbound
4/r
Black Dot:structured mesh point excluded from SPH list
Blue Circle:Boundary Polygon Red Circle:
structured mesh point Included in SPH list
Gives a reasonablyWell spaced grid
Easy to identify “node Clouds”
IDEA stolen from Immersed BoundaryMethods of Fotis
Sotiropoulos
h = r
ResultsRadial Movement of Front with Time
Fine grid radial symmetry solution
Melt pattern atAn intermediate times
Also works when points are“Jostled”
Unstructured Mesh
Delaunay
“Patchy”
Good Results
But sensitiveto choice of h
P
So far very basic calculations
But they show promise—
Need to look at
Adaptivity
Lagrangian
Fans Toes Shoreline
Two Sedimentary Moving Boundary Problems of Interest
Moving Boundaries in Sediment Transport
1km
Examples of Sediment FansMoving Boundary
How does sediment-basement interfaceevolve
Badwater Deathvalley
An Ocean Basin
Melting vs. Shoreline movement
h(x,y,t)
q
bed-rock
ocean
y
shoreline
x = s(t)
land surface
(x,y,t)
A 2-D Front -Limit of Cliff face Shorefront But Account of Subsidence and relative ocean level
0hif),t,y,x(LhH
)h(tH
Enthalpy Sol.
xy
]L/H,[MINfrac 1
Solve on fixed gridin plan view
Track Boundary by calculating in each cell
A 2-D problem Sediment input into an oceanwith an evolving trench driven By hinged subsidence
With Trench
Need to account for Interaction with channels whichcan avulse
Increasing Flexibility
GRID---Structure
Easy ApproximationEfficient solution Ax=b —
Difficult to Fit Geometry
MESH---FEM
More Complex approximationLess Efficient solution Ax=b —
Easy to Fit GeometryCould be Difficult to adapt
“CLOUD”---SPH
“Poor” approximationEven Less Efficient solution Ax=b —
Very Easy to Fit GeometryEasy to adapt