Groundwater Modeling: Finite Element Method

Wolfgang Kinzelbach
Institute of Environmental Engineering

ETH Zrich
Groundwater modeling:
Finite element method
Groundwater II
Finite element method
using2D flowas an example
) , ( ) ( ) , , (
1
y x w t h t y x h
i
n
i
i
=
=
) , ( y x w
i
Interpolation functions
patch
1
Nodefunctions
Nodal valuein steadystatecase
h
i
(t)
h
i
Method comprises discretisation of domain interpolation
between nodal values
Trial solution
Simplest variant: Triangular elementsand
linear interpolation
h
2
h
1
h
3
Element
Node 1
(x
1
,y
1
)
Node 2
(x
2
,y
2
)
Node 3
(x
3
,y
3
)
h
x
y
Comparemethodfor constructionof
contourlinesof heads!
Triangular elementsand linear interpolation
[ ]
1 1 1 1 2 2 2 2 3 3 3 3
1 1
2 2
3 3
( , )
1
( ) ( ) ( )
1
with D 1 2F
1
h x y x y
h A B x C y h A B x C y h A B x C y
D
x y
x y
x y

= + +
= + + + + + + + +
= =
1 2 3 2 1 3 2 1 2 1 3
3 1 2 1 3 2 1 3 1 3 2
2 3 1 3 2 1 3 2 3 2 1
x x C y y B x y y x A
= = =
= = =
= = =
Calculationof Darcy-velocity
) (
1
) (
1
3 3 2 2 1 1
3 3 2 2 1 1
h C h C h C
D
k v
h B h B h B
D
k v
f y
f x
+ + =
+ + =
Q
S1
=orthogonal projection
of v on side32 thickness
length32
2
3 2
2
3 2
3 2
3 2
32
) y y ( ) x x ( /
) x x (
y y
n +
Fluxacrosstriangleside32
1
3
2
n
32
v
Unit normal vector of side32:
Calculationof fluxesacrosstriangle
sides
m
x x
y y
v
v
Q
y
x
S

=
) (
3 2
3 2
1
m
x x
y y
v
v
Q
y
x
S

=
) (
1 3
1 3
2
m
x x
y y
v
v
Q
y
x
S

=
) (
2 1
2 1
3
: Q
S1
+Q
S2
+Q
S3
=0
Continuity requires:
Concept of nodal fluxes
1 2 3
3 1 2
2 3 1
5 . 0 5 . 0
5 . 0 5 . 0
5 . 0 5 . 0
S S
S S
S S
Q Q W
Q Q W
Q Q W
+ =
+ =
+ =
W
1
=-0.5Q
S1
W
2
=-0.5Q
S2
W
3
=-0.5Q
S3
) ( ) ( ) ( (
2
) ( ) ( ) ( (
2
) ( ) ( ) ( (
2
3 3 3 3 3 2 3 2 3 2 1 3 1 3 1 3
2 3 2 3 3 2 2 2 2 2 1 2 1 2 1 2
1 3 1 3 3 1 2 1 2 2 1 1 1 1 1 1
C C B B h C C B B h C C B B h
D
T
W
D
T
W
D
T
W
+ + + + + =
+ + + + + =
+ + + + + =
Flux across side is distributed half-half over endpoints
Concept of nodal fluxes
Nodal flux contributions of element e to node i
e
i
Same fromall other elements in the patch.
Summation leads to water balance. But before we can
do that, a unique notation is required .
D T C C B B E
j i j i ij
2 / ) ( + =
j
j
ij i
h E W

=
=
3
1
Transformation intoglobal
matrices
K=n(e,i) (i=1,2,3; e=1,...M)
) , ( ); , ( , ,
, ,
j e n l i e n k E E W W
h h y y x x
e
kl
e
ij
e
k
e
i
k
e
i k
e
i k
e
i
= =

Simplifiednotation
Definition:
Incidence matrix
Embeddingof 3 x 3 element matrix
intoglobal matrix
n(e,3)
n(e,3)
n(e,2)
n(e,2)
n(e,1)
n(e,1)
N
N
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0

E E E
E E E
E E E
0 ) )( (
1 1
= + = +

= =
k l
M
e
e
kl
l
k
M
e
e
k
Q h E Q W
Water balancethroughaddingup of
nodal fluxesin global form
Sumis over all elements of patch k, k= 1 to n
Nodal fluxes + source/sink in node = 0
Node k
e
5
e
4
e
3
e
2
e
1
In example: M=5
Procedureat boundaries
Prescribed piezometric head at node i:
Replace balance equation at node i by h
i
(t)=f(t)
Prescribed boundary flux:
Distribute over boundary nodes and take as nodal source/sink
Boundaryflux
Procedurefor areal recharge
Areal flux is divided into three parts and added to nodal fluxes
Water balancethroughaddingup of
nodal fluxesin global form
Sum represents water balance over colored part of patch
Without areally
distributedrecharge
Withareally
distributedrecharge
Finally: Equationsystem
[ ][ ] [ ] b h A =
Differences to FD: larger bandwidth, more non-zero coefficients,
different weights in matrix A
Advantage: local refinement feasible, flexible adaptation to
complex aquifer geometry, easy implementation of anisotropy
Transient case
Difference to FD: Non-local storage
FD: Change of head in node i contributes only to storage in
cell i
FE: Change of head in node i contributes to storage at all
other nodes of the patch
h
FD
FE
h
Transient case
Consequence: Storage matrix in FE non-diagonal

= =
= +
N
j
i j j i
j
N
j
j i
B h A
t
h
R
1
,
1
,
This equation is usually solved with finite differences in time
Element contribution to storage in node i
1
of element e
) 2 (
4
1
3
1
3 2 1
t
h
t
h
t
h
F S
i i i
e e
F
e
element area
h
i1
h
i2
h
i3
FE: GalerkinMethod
The derivation of FE described above only works for
triangular elements with linear interpolation function. A
general method to arrive at finite element equations for
arbitrary elements with arbitrary interpolation functions is
the Galerkinmethod
Steps
Trial solution with unknown pivot values and interpolation
function
Insertion into flow equation leads to residual
Residual is minimized by optimal choice of unknown pivot values
at nodes
Minimization uses Galerkinprinciple
(Weighting function =Interpolation function)
Example: steadystateflowin 1D
=
=
n i
i i
x w h x h
, 1
) ( ) (
w
i
arehat functions
Flowequation(symbolically)
0 )) ( ( = x h L
Insert trial solution: ) ( )) (
( x x h L =
Evenlydistributedminimizationof error
(n conditionsfor thesmallnessof residual as n headsareunknown)
( ) ( ) 0 1,...,
i
x w x dx i n = =
i i+1 i-1
1
0
Transfer of onederivativeto weightingfunction, as
interpolation function is linear (2nd derivative 0)
Avoids -functions
2
2
( ) 0 1,...,
j j i
T h w q w x dx i n
x

= =

( ) ( ) 0 1,...,
j j i i
T h w w x dx qw x dx i n
x x

= =

Conditionsleadto n linear equationsfor n unknownheadsh
i
( )( ) ( ) ,...,
ij j i
a h b i 1 n = =
1,
( , ) ( , )
i i
i n
h x y h w x y
=
=

( ( , )) 0 L h x y
w
i
arepyramidfunctions
Flowequation(symbolically)
=
Insert trial solution:
( ( , )) ( , ) L h x y x y =
Triangular elementswithlinear interpolationfunctions
Evenlydistributedminimizationof error
(n conditionsfor thesmallnessof residual as n headsareunknown)
( , ) ( , ) 0 1,...,
i
x y w x y dxdy i n = =
1
0
Transfer of onenabla-operator to weightingfunction, as
interpolation function is linear (2nd derivative 0)
Avoids -functions
( )
( )
( , ) 0 1,...,
j j i
T h w q w x y dx dy i n = =
Conditionsleadto n linear equationsfor n unknownheadsh

i
Advantage of Galerkinformalism: Canbedonein the
sameway in all dimensions. 2D showsthat anisotropy
canbeincorporatednaturally. J ust makeT a tensor.
( )( ) ( ) ,...,
ij j i
a h b i 1 n = =
Transient flowin GalerkinFE
Startingout from:
1,
( , , ) ( ) ( , )
i i
i n
h x y t h t w x y
=
=

leadsto onemoretermin final set of equation
( )( ) ( )( ) ( ) ,...,
j
ij ij j i
dh
r a h b i 1 n
dt
+ = =
Take finite differencesfor thediscretizationof thetime
derivativetermand useh(t+t) in second term.
( )
( ( ) ( , )) ( ) ( , ) ( , ) 0
1,...,
j j j j i
d
S h t w x y T h t w x y q w x y dx dy
dt
i n

+ =

=

ApplyingGalerkinformalism
Transient flowin 2D: Details (1)
( )
1 1 1
1 1 1
( , ) 0 1,...,
)) 0
( ( ) ( ) (
) )
( ) (
(
j j j j i
n n N
j j i
j j j j
j j j
n n N
i i
j j i i
j j j j
j j J
j
i
j
T h w q S h w w x y dxdy i n
t
h w wdxdy
T h w T h w q S
x x y y t
w w
h w w qw dxdy
T h w T h w S
x x y y t
w x
w T
w y
= = =
= = =

+ = =

=
+ +

+ +

1
) 0
n
j n i
j
n h ds Boundary integral q wds
=

= =

i
1 1
1
0
( )
n n
j
ij j ij i
j j
j j
i i
ij
ij i j
i i n i
M
e
i i
e
h
Ph R F
t
with
w w
w w
P T T dxdy
x x y y
R Sww dxdy
F qwdxdy q wds
Nowuse w
= =
+ =

= +

=
= +
=


( )
, ( 0)
j j
i i
ij
ij i j
i i i
e
e e
e e
e e e
e
e e e e
e
e e e
n
e
P T T dxdy
x x y y
R S dxdy
F q dxdy q ds
for element e containing nodes i j otherwise

= +

=
= +

2 2
( )( ) ( )( )
( )
( ) ( ) 2 2
0
2 ,1
(( ) ( ) (
ij
ij
i
e e
j k k i k j k i e e e e e
i j i j
e e
i k i k
e e e e e
ij ij
e
j k k j j k w k
for i j
y y y y x x x x
T T
P B B C C
y y x x D D
for i j
for i or j not in e
R S D with
otherwise for i j for i j
F Q x y x y y y x x x

+

= = +

+

=

= =

=

= + + ) )/ 0.5
, ,
,
e
j w n
w w
n
y D Lq
with x y coordinates of well with pumping rateQ
L length of boundary edge q fluxover that edge
+
Example for Matlab-Code fem_simple.m
(0,0)
(100,0)
(0,100)
(100,100)
coordinates in m
y
x
1
3
1
5
2
4
Nodal number
Element number
1
1
4
3
2
Pumping rate in
node 3: 1 m
3
/s
Transmissivity
of elements:
0.1 m
2
/s
Nodes 1 and 2
fixed heads at 50 m
Limmat
Sihl
Example for more
professional application
measured head contours
computed head contours
streamlines
0.00 - 0.05
0.05 - 0.10
0.10 - 0.15
0.15 - 0.20
0.20 - 0.25
0.25 - 0.30
0.30 - 0.35
0.35 - 0.40
0.40 - 0.45
0.45 - 0.50
0.50 - 0.55
0.55 - 0.60
0.60 - 0.65
0.65 - 0.70
0.70 - 0.75
0.75 - 0.80
0.80 - 0.85
0.85 - 0.90
0.90 - 0.95
0.95 - 1.00
1.00 - 1.05
Computed proportions of
city water

Groundwater Modeling: Finite Element Method

Uploaded by

Copyright:

Available Formats

Groundwater Modeling: Finite Element Method

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Groundwater Modeling: Finite Element Method

Uploaded by

Copyright:

Available Formats

Wolfgang Kinzelbach

Institute of Environmental Engineering

Conditionsleadto n linear equationsfor n unknownheadsh

Transient flowin 2D: Details (3)

You might also like