Professional Documents
Culture Documents
Agern All 5
DK-2970 Hrsholm
Denmark
Tel:
+45 4516 9200
Support: +45 4516 9333
Fax:
+45 4516 9292
mikebydhi@dhigroup.com
www.mikebydhi.com
PLEASE NOTE
COPYRIGHT
LIMITED LIABILITY
PRINTING HISTORY
2010-07-07/MIKE_321_FM_SCIENTIFIC_DOC.DOCX/OSP/ORS/Manuals.lsm
CONTENTS
MIKE 21 & MIKE 3 FLOW MODEL FM
Hydrodynamic and Transport Module
Scientific Documentation
INTRODUCTION .............................................................................................. 1
2.1
2.1.1
2.1.2
2.1.3
2.1.4
2.1.5
2.2
2.3
2.3.1
2.3.2
2.3.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.10.1
2.10.2
2.10.3
2.10.4
3.1
3.1.1
3.1.2
3.1.3
3.2
3.3
3.3.1
3.3.2
3.3.3
VALIDATION ................................................................................................. 43
4.1
4.1.1
4.1.2
4.1.3
REFERENCES ................................................................................................. 49
ii
Introduction
INTRODUCTION
This document presents the scientific background for the new MIKE
21 & MIKE 3 Flow Model FM1 modelling system developed by DHI
Water & Environment. The objective is to provide the user with a
detailed description of the flow and transport model equations,
numerical discretization and solution methods. Also model validation
is discussed in this document.
MIKE 21 & MIKE 3 Flow Model FM is based on a flexible mesh
approach and it has been developed for applications within
oceanographic, coastal and estuarine environments. The modelling
system may also be applied for studies of overland flooding.
The system is based on the numerical solution of the two/threedimensional incompressible Reynolds averaged Navier-Stokes
equations invoking the assumptions of Boussinesq and of hydrostatic
pressure. Thus, the model consists of continuity, momentum,
temperature, salinity and density equations and it is closed by a
turbulent closure scheme. For the 3D model the free surface is taken
into account using a sigma-coordinate transformation approach.
The spatial discretization of the primitive equations is performed using
a cell-centred finite volume method. The spatial domain is discretized
by subdivision of the continuum into non-overlapping elements/cells.
In the horizontal plane an unstructured grid is used while in the
vertical domain in the 3D model a structured mesh is used. In the 2D
model the elements can be triangles or quadrilateral elements. In the
3D model the elements can be prisms or bricks whose horizontal faces
are triangles and quadrilateral elements, respectively.
Including the MIKE 21 Flow Model FM (two-dimensional flow) and MIKE 3 Flow Model FM (three-dimensional
flow)
Scientific Documentation
Governing Equations
GOVERNING EQUATIONS
2.1
2.1.1
S
x y z
(2.1)
and the two horizontal momentum equations for the x- and ycomponent, respectively
u u 2 vu wu
1 pa
fv g
t x
y
z
x 0 x
g
0 z
1 sxx sxy
u
dz
Fu t
us S
x
0h x
y
z z
v v 2 uv wv
1 pa
fu g
t y
x
z
y 0 y
g
0 z
1 s yx s yy
v
dz
Fv t vs S
y
0h x
y
z z
(2.2)
(2.3)
Scientific Documentation
Fu
u u v
2 A A
x
x y y x
(2.4)
Fv
u v v
2A
A
x y x y y
(2.5)
1
u v
u
v
w 0, ,
sx , sy
t
x
y
z z 0 t
(2.6)
At z d :
d
d
1
u v
u
v
w 0, ,
bx , by
x
y
z z 0 t
(2.7)
hS P E
t
x
y
(2.8)
hu udz ,
d
hv vdz
d
(2.9)
(T , s)
(2.10)
Governing Equations
FT Dv
H Ts S
t
x
y
z
z
z
(2.11)
s us vs ws
s
Fs Dv s s S
t x
y
z
z
z
(2.12)
Dh Dh T , s
x y
y
x
FT , F s
(2.13)
and Dv
t
T
(2.14)
Qn
T
T p P Te E
z 0 c p
(2.15)
At z d :
T
0
z
Scientific Documentation
(2.16)
s
0
z
At z d :
(2.18)
s
0
z
q
qv 0
v
E 0 lv
0
qv 0
(2.19)
where q v is the latent heat flux and lv 2.5 10 6 is the latent heat of
vaporisation of water.
2.1.3
FC Dv
k p C Cs S
t
x
y
z
z
z
(2.20)
FC Dh Dh C
x y
y
x
(2.21)
Governing Equations
2.1.4
Turbulence model
The turbulence is modelled using an eddy viscosity concept. The eddy
viscosity is often described separately for the vertical and the
horizontal transport. Here several turbulence models can be applied: a
constant viscosity, a vertically parabolic viscosity and a standard k-
model (Rodi, 1984). In many numerical simulations the small-scale
turbulence can not be resolved with the chosen spatial resolution. This
kind of turbulence can be approximated using sub-grid scale models.
Vertical eddy viscosity
The eddy viscosity derived from the log-law is calculated by
2
zd
z d
t U h c1
c2
h
h
(2.22)
t t* (1 aRi ) b
(2.23)
(2.24)
t c
Scientific Documentation
k2
(2.25)
where k is the turbulent kinetic energy per unit mass (TKE), is the
dissipation of TKE and c is an empirical constant.
The turbulent kinetic energy, k, and the dissipation of TKE, , are
obtained from the following transport equations
k uk vk wk
k
P B
Fk t
t
x
y
z
z k z
(2.26)
u v w
t
x
y
z
c1 P c3 B c2
F t
z z k
(2.27)
u 2 v 2
xz u yz v
P
t
z z
0 z 0 z
t 2
N
t
(2.28)
(2.29)
g
0 z
(2.30)
( Fk , F ) Dh Dh (k , )
x y
y
x
(2.31)
Governing Equations
Table 2.1
c1
1.44
0.09
c 2
1.92
c3
0
0.9
1.0
1.3
At the surface the boundary conditions for the turbulent kinetic energy
and its rate of dissipation depend on the wind shear, Us
At z = :
k
1
U2s
c
(2.32)
U3s
for U s 0
zb
k
0
z
ah
3/ 2
for Us 0
(2.33)
1
c
2
b
U3b
(2.34)
zb
Scientific Documentation
(2.35)
1 u u j
S ij i
2 x j xi
2.1.5
(i, j 1,2)
(2.36)
z zb
, x x,
h
y y
(2.37)
where varies between 0 at the bottom and 1 at the surface. The coordinate transformation implies relations such as
1
z h
(2.38)
1 d
h 1 d
h
,
,
x y h y
y
x y x h x
(2.39)
hS
t x y
(2.40)
hu hu 2 hvu hu
h pa
fvh gh
t
x
y
x 0 x
hg
1 s
dz xx xy hFu
x
0 x
y
v u
hus S
h
hv huv hv 2 hv
h pa
fuh gh
t
x
y
y 0 y
hg
1 s
v v
dz yx yy hFv
hvs S
y
0 x
y
h
hT huT hvT hT
t
x
y
(2.42)
(2.43)
Dv T
hFT
hH hTs S
h
hs hus hvs hs
hFs
t
x
y
10
(2.41)
Dv s
hs s S
h
(2.44)
Governing Equations
hk huk hvk hk
t
x
y
1
hFk
h
t k
h( P B )
k
(2.45)
h hu hv h
t
x
y
hF
1
h
h c1 P c3 B c2
hC huC hvC hC
Dv C
hFC
hk p C hC s S
t
x
y
(2.46)
(2.47)
h
1
d
d
h
h
v
u
v
w u
h
x
y
x
y
t
(2.48)
hFu
u u v
2hA hA
x
x y y x
(2.49)
hFv
u v
v
hA 2hA
x y x y
y
(2.50)
h( FT , Fs , Fk , F , Fc )
hDh hDh (T , s, k , , C )
x y
y
x
(2.51)
The boundary condition at the free surface and at the bottom are given
as follows
At =1:
h
u v
0,
,
sx , sy
0 t
At =0:
Scientific Documentation
(2.52)
(2.53)
11
h
u v
,
bx , by
0 t
0,
2.2
d
dt
vR
d
dt
(2.54)
h
1 hu hv cos h
hS
t R cos
(2.55)
hu
1 hu 2 hvu cos hu
u
f tan vh
t
R cos
1 1 pa g
gh
R cos 0 0
hFu
1 s
dz xx cos xy
(2.56)
v u
hus S
h
hv
1 huv hv 2 cos hv
u
f tan uh
t
R cos
1 1 pa g
gh
R 0 0
hFv
1 1 s yx s yy
dz
0 cos
v v
hvs S
h
hT
1 huT hvT cos hT
t
R cos
Dv T
hFT
hH hTs S
h
12
(2.57)
(2.58)
Governing Equations
hs
1 hus hvs cos hs
t R cos
hFs
Dv s
hss S
h
hk
1 huk hvk cos hk
t
R cos
1 t k
h( P B )
hFk
h k
h
1 hu hv cos h
t
R cos
1
hF
h
h c1 P c3 B c2
k
hC
1 huC hvC cos hC
t
R cos
Dv C
hFC
hk pC hCs S
h
(2.59)
(2.60)
(2.61)
(2.62)
h
1
u d v d
u h v h
w
h
R cos R y
t R cos R
(2.63)
h
1 hu hv cos
hS
t R cos
2.3
2.3.1
(2.64)
hS
t
x
y
Scientific Documentation
(2.65)
13
hu hu 2 hvu
h pa
fvh gh
t
x
y
x 0 x
gh 2 sx bx 1 s xx sxy
2 0 x 0 0 0 x
y
(2.66)
fuh gh
t
x
y
y 0 y
gh 2 sy by 1 s yx s yy
2 0 y 0 0 0 x
y
x
y
(2.67)
hu udz ,
d
hv vdz
(2.68)
The lateral stresses Tij include viscous friction, turbulent friction and
differential advection. They are estimated using an eddy viscosity
formulation based on of the depth average velocity gradients
Txx 2 A
2.3.2
u v
u
v
, Txy A
, T yy 2 A
x
y
y x
(2.69)
hT hu T hv T
hFT hH hTs S
t
x
y
(2.70)
hs hu s hv s
hFs hs s S
t
x
y
(2.71)
14
Governing Equations
2.3.3
hFC hk p C hCs S
t
x
y
(2.72)
2.4
d
dt
vR
d
dt
(2.73)
h
1 hu hv cos
t R cos
(2.74)
hu
1 hu 2 hvu cos
u
f tan vh
t
R cos
s
1 h pa gh 2 1 s xx
cos xy
gh
R cos 0 2 0 0
sx bx
(2.75)
hv
1 huv hv 2 cos
u
f tan uh
t
R cos
1 h pa gh 2 1 1 s yx s yy
gh
R 0 2 0 0 cos
(2.76)
sy by
hT
1 hu T hv T cos
hFT hH hTs S
t
R cos
Scientific Documentation
(2.77)
15
hs
1 hu s hv s cos
hFs hs s S
t
R cos
hC
1 hu C hv C cos
hFC hk p C hCs S
t
R cos
2.5
(2.78)
(2.79)
Bottom Stress
b
c f ub ub
0
(2.80)
U b c f ub
(2.81)
cf
g
C2
(2.82)
Mh
(2.83)
1/ 6 2
16
1
1 z b
ln
z
0
(2.84)
Governing Equations
where =0.4 is the von Krmn constant and z 0 is the bed roughness
length scale. When the boundary surface is rough, z 0 , depends on the
roughness height, k s
z 0 mk s
(2.85)
2.6
25.4
k s1 / 6
(2.86)
Wind Stress
s a cd u w u w
(2.87)
a c f uw
0
(2.88)
c a
w10 wa
c ca
w10 wa wa w10 wb
c f c a b
wb wa
cb
w10 wb
(2.89)
where ca, cb, wa and wb are empirical factors and w10 is the wind
velocity 10 m above the sea surface. The default values for the
empirical factors are ca = 1.25510-3, cb = 2.42510-3, wa = 7 m/s and
wb = 25 m/s. These give generally good results for open sea
applications. Field measurements of the drag coefficient collected over
Scientific Documentation
17
lakes indicate that the drag coefficient is larger than open ocean data.
For a detailed description of the drag coefficient see Geernaert and
Plant (1990).
2.7
Ice Coverage
It is possible to take into account the effects of ice coverage on the
flow field.
In areas where the sea is covered by ice the wind stress is excluded.
Instead, the surface stress is caused by the ice roughness. The surface
s
c f us us
0
(2.90)
U s c f u s
(2.91)
Mh
(2.92)
1/ 6 2
25.4
k s1 / 6
(2.93)
18
1
1 z s
ln
z
0
(2.94)
Governing Equations
where =0.4 is the von Krmn constant and z 0 is the bed roughness
length scale. When the boundary surface is rough, z 0 , depends on the
roughness height, k s
z 0 mk s
(2.95)
2.8
Tidal Potential
The tidal potential is a force, generated by the variations in gravity due
to the relative motion of the earth, the moon and the sun that act
throughout the computational domain. The forcing is expanded in
frequency space and the potential considered as the sum of a number
of terms each representing different tidal constituents. The forcing is
implemented as a so-called equilibrium tide, which can be seen as the
elevation that theoretically would occur, provided the earth was
covered with water. The forcing enters the momentum equations (e.g.
(2.66) or (2.75)) as an additional term representing the gradient of the
equilibrium tidal elevations, such that the elevation can be seen as
the sum of the actual elevation and the equilibrium tidal potential.
ACTUAL T
(2.96)
T ei H i fi Li cos(2
i
t
bi i0 x)
Ti
(2.97)
(2.98)
Scientific Documentation
19
Table 2.2
277.02+481267.89T+0.0011T
280.19+36000.77T+0.0003T
334.39+4069.04T+0.0103T
259.16+1934.14T+0.0021T
Longitude of perihelion
ps
281.22+1.72T+0.0005T
L 3sin 2 ( y) 1
L sin(2 y)
L cos2 ( y)
2.9
fi
ui
Mm
Mf
-23.7 sin(N)
Q1, O1
10.8 sin(N)
K1
-8.9 sin(N)
2N2, 2, 2, N2, M2
-2.1 sin(N)
K2
-17.7 sin(N)
Wave Radiation
The second order stresses due to breaking of short period waves can be
included in the simulation. The radiation stresses act as driving forces
for the mean flow and can be used to calculate wave induced flow. For
3D simulations a simple approach is used. Here a uniform variation is
used for the vertical variation in radiation stress.
20
Governing Equations
2.10
Heat Exchange
The heat exchange with the atmosphere is calculated on basis of the
four physical processes
Latent and sensible heat fluxes and long-wave radiation are assumed
to occur at the surface. The absorption profile for the short-wave flux
is approximated using Beers law. The attenuation of the light intensity
is described through the modified Beer's law as
I (d ) 1 I 0 e d
(2.99)
(2.100)
( z )
q sr ,net 1
0c p
e ( z )
(2.101)
(2.102)
The calculation of the latent heat flux, sensible heat flux, net short
wave radiation, and net long wave radiation as described in the
following sections.
In areas covered by ice the heat exchange is excluded.
Scientific Documentation
21
2.10.1 Vaporisation
Daltons law yields the following relationship for the vaporative heat
loss (or latent flux), see Sahlberg, 1984
(2.103)
0.2167
ei
Ti Tk
(2.104)
in which subscript i refers to both water and air. The vapour pressure
close to the sea, ewater , can be expressed in terms of the water
temperature assuming that the air close to the surface is saturated and
has the same temperature as the water
1
1
ewater 6.11e K
Tk Twater Tk
(2.105)
1
1
eair R 6.11e K
Tk Tair Tk
(2.106)
Replacing Qwater and Qair with these expressions the latent heat can
be written as
22
Governing Equations
qv Pv a1 bW
1 2m
1
1
exp K
Tk Twater Tk
Twater Tk
1
R exp K T T T
air
k
Tair Tk
(2.107)
(2.108)
Scientific Documentation
23
r
E0 0 1.000110 0.034221cos() 0.001280 sin()
r
0.000719 cos(2) 0.000077 sin(2)
(2.109)
where r0 is the mean distance to the sun, r is the actual distance and
the day angle (rad ) is defined by
2 (d n 1)
365
(2.110)
(2.111)
24
(2.112)
and the sunrise angle, sr (rad ) , and the sunset angle ss (rad ) are
(2.113)
24
24
(2.114)
Governing Equations
(2.115)
H
q0 a3 b3 cos i
qs
H0
(2.116)
where
(2.117)
(2.118)
E
4
12 t displacement LS L E t t local
12
60
60
(2.119)
(2.120)
Scientific Documentation
25
(2.121)
tan 2 (i r )
2 sin (i r ) tan 2 (i r )
(2.122)
30 altitude
0.48 0.05 5 altitude 30
25
altitude 30
0.05
(2.123)
altitude 90
arccos(sin( ) sin( ) cos( ) cos( ) cos( i ))
(2.124)
10 6
3600
(2.125)
26
Governing Equations
atmosphere and the sea surface. The long wave emittance from the
surface to the atmosphere minus the long wave radiation from the
atmosphere to the sea surface is called the net long wave radiation and
is dependent on the cloudiness, the air temperature, the vapour
pressure in the air and the relative humidity. The net outgoing long
wave radiation, qlr ,net (W / m 2 ) , is given by Brunts equation (See
Lind and Falkenmark, 1972)
(2.126)
(2.127)
ed 10 R esaturated
(2.128)
7.38 10
Scientific Documentation
(2.129)
27
28
Numerical Solution
NUMERICAL SOLUTION
3.1
Spatial Discretization
The discretization in solution domain is performed using a finite
volume method. The spatial domain is discretized by subdivision of the
continuum into non-overlapping cells/elements.
In the two-dimensional case the elements can be arbitrarily shaped
polygons, however, here only triangles and quadrilateral elements are
considered.
In the three-dimensional case a layered mesh is used: in the horizontal
domain an unstructured mesh is used while in the vertical domain a
structured mesh is used (see Figure 3.1gure 3.1). The vertical mesh is
based on either sigma coordinates or combined sigma/z-level
coordinates. For the hybrid sigma/z-level mesh sigma coordinates are
used from the free surface to a specified depth and z-level coordinates
are used below. The different types of vertical mesh are illustrated in
Figure 3.2. The elements in the sigma domain and the z-level domain
can be prisms with either a 3-sided or 4-sided polygonal base. Hence,
the horizontal faces are either triangles or quadrilateral element. The
elements are perfectly vertical and all layers have identical topology.
Figure 3.1
Scientific Documentation
29
Figure 3.2
30
Numerical Solution
3.1.1
Vertical Mesh
For the vertical discretization both a standard sigma mesh and a
combined sigma/z-level mesh can be used. For the hybrid sigma/zlevel mesh sigma coordinates are used from the free surface to a
specified depth, z, and z-level coordinates are used below. At least
one sigma layer is needed to allow changes in the surface elevation.
Sigma
In the sigma domain a constant number of layers, N, are used and
each sigma layer is a fixed fraction of the total depth of the sigma
layer, h, where = max( , ). The discretization in the sigma
domain is given by a number of discrete -levels , = 1, ( +
1) . Here varies from 1 = 0 at the bottom interface of the lowest
sigma layer to +1 = 1 at the free surface.
Variable sigma coordinates can be obtained using a discrete
formulation of the general vertical coordinate (s-coordinate) system
proposed by Song and Haidvogel (1994). First an equidistant
discretization in a s-coordinate system (-1 s 0) is defined
=
+ 1
= 1, ( + 1)
(3.1)
i 1 c si 1 c csi i 1, N 1
(3.2)
where
= 1
sinh
+
sinh
1
tanh + 2
tanh
(2 )
(3.3)
2tanh
(2 )
Scientific Documentation
31
Figure 3.3
Figure 3.4
Combined sigma/z-level
In the z-level domain the discretization is given by a number of
discrete z-levels , = 1, ( + 1) , where Nz is the number of layers
in the z-level domain. z1 is the minimum z-level and +1 is the
maximum z-level, which is equal to the sigma depth, z. The
corresponding layer thickness is given by
= +1
32
= 1,
(3.4)
Numerical Solution
(3.5)
(2 )
1
(3.6)
= 1,
(3.7)
= 1
(+1 )
,
>
< < +1 1
(3.8)
Scientific Documentation
33
3.1.2
Figure 3.5
Figure 3.6
(3.9)
34
Numerical Solution
S
t
x
y
(3.10)
h
U hu ,
hv
hu
1
u
Fx I hu 2 g (h 2 d 2 ) , FxV hA 2
huv
u v
hA
y x
hv
V
Fy I hvu
, Fy hA y x
2 1
2
2
hv
g
(
h
d
)
v
2
a
hA 2
(3.11)
2
d
h pa gh 1 s xx sxy
g
fvh
0 x 2 0 x 0 x
y
x
S
sx bx hus
0 0
2
g d fuh h pa gh 1 s yx s yy
y
0 y 2 0 y 0 x
y
sy by
hvs
0 0
Scientific Documentation
35
I
V
U FxI Fy FI FxV Fy FV
S
t
x
y
x
y
(3.12)
hu
Fx I hu 2 g (h 2 d 2 ) , FxV hA 2
2
x
huv
u v
hA
y x
hv
V
Fy hvu
, Fy
hv 2 1 g (h 2 d 2 )
2
h
F h u , F V
h v
I
t
h
t
h
0
d
h
S g
fvh
0
x
g d fuh h
y
0
u v
hA
y x
v
hA 2
(3.13)
u
v
pa hg
1 s
dz xx
z
x 0
x
0 x
pa hg
1 s
dz yx
y 0 z y
0 x
s
xy hus
y
s
yy hvs
Integrating Eq. (3.9) over the ith cell and using Gausss theorem to
rewrite the flux integral gives
36
U
d ( F n) ds S (U )d
Ai t
i
Ai
(3.14)
Numerical Solution
U i 1
t
Ai
NS
F n
Si
(3.15)
Scientific Documentation
37
3.1.3
Transport equations
The transport equations arise in the salt and temperature model, the
turbulence model and the generic transport model. They all share the
form of Equation Eq. (2.20) in Cartesian coordinates. For the 2D case
the integral form of the transport equation can be given by Eq. (3.9)
where
U hC
F I huC , hvC
C
C
F V hDh
, hDh
x
y
(3.16)
S hk pC hCs S .
For the 3D case the integral form of the transport equation can be
given by Eq. (3.9) where
U hC
F I huC , hvC , hC
C
C
D C
F V hDh
, hDh
, h h
x
y
h
(3.17)
S hk pC hCs S .
38
Numerical Solution
3.2
Time Integration
Consider the general form of the equations
U
G U
t
(3.18)
For 2D simulations, there are two methods of time integration for both
the shallow water equations and the transport equations: A low order
method and a higher order method. The low order method is a first
order explicit Euler method
Un1 Un t G(Un )
(3.19)
where t is the time step interval. The higher order method uses a
second order Runge Kutta method on the form:
U n 1 U n 12 t G (U n )
2
U n 1 U n t G (U n 1 )
(3.20)
(3.21)
(3.22)
The horizontal terms are integrated using a first order explicit Euler
method and the vertical terms using a second order implicit trapezoidal
rule. The higher order method can be written
Scientific Documentation
39
Un 1 2 14 t Gv (Un 1 2 ) Gv (U n ) U n 12 t Gh (U n )
Un 1 12 t Gv (U n 1 ) Gv (U n ) U n t Gh (U n 1 2 )
(3.23)
The horizontal terms are integrated using a second order Runge Kutta
method and the vertical terms using a second order implicit trapezoidal
rule.
The low order method used for the 3D transport equation can written
as
Un1 12 t GvV (Un1 ) GvV (Un ) Un t Gh (Un ) t GvI (Un )
(3.24)
The horizontal terms and the vertical convective terms are integrated
using a first order explicit Euler method and the vertical viscous terms
are integrated using a second order implicit trapezoidal rule. The
higher order method can be written
U n 1 2 14 t GvV (U n 1 2 ) GvV (U n )
U n 12 t Gh (U n ) 12 t G vI (U n )
U n 1 12 t GvV (U n 1 ) GvV (U n )
(3.25)
U n t Gh (U n 1 2 ) t GvI (U n 1/ 2 )
The horizontal terms and the vertical convective terms are integrated
using a second order Runge Kutta method and the vertical terms are
integrated using a second order implicit trapezoidal rule for the
vertical terms.
3.3
Boundary Conditions
3.3.1
Closed boundaries
Along closed boundaries (land boundaries) normal fluxes are forced to
zero for all variables. For the momentum equations this leads to fullslip along land boundaries.
3.3.2
Open boundaries
The open boundary conditions can be specified either in form of a unit
discharge or as the surface elevation for the hydrodynamic equations.
For transport equations either a specified value or a specified gradient
can be given.
40
Numerical Solution
3.3.3
An element is wet if the water depth is greater than hwet . Both the
mass fluxes and the momentum fluxes are calculated.
The wetting depth, hwet , must be larger than the drying depth, hdry ,
and flooding depth, h flood , must satisfy
hdry h flood hwet
(3.26)
The default values are hdry 0.005 m , h flood 0.05 m and hwet 0.1m .
Note, that for very small values of the tolerance depth, hwet ,
unrealistically high flow velocities can occur in the simulation and
give cause to stability problems.
Scientific Documentation
41
42
Validation
VALIDATION
The new finite-volume model has been successfully tested in a number
of basic, idealised situations for which computed results can be
compared with analytical solutions or information from the literature.
The model has also been applied and tested in more natural
geophysical conditions; ocean scale, inner shelves, estuaries, lakes and
overland, which are more realistic and complicated than academic and
laboratory tests. A detailed validation report is under preparation.
This chapter presents a comparison between numerical model results
and laboratory measurements for a dam-break flow in an L-shaped
channel.
Additional information on model validation and applications can be
found here
http://mikebydhi.com/Download/DocumentsAndTools/PapersAndDocs.aspx
4.1
Physical experiments
A comprehensive experimental study of a dam-break flow in a channel
with a 90 bend has been reported by Frazo and Zech (2002, 1999a,
1999b). The channel is made of a 3.92 and a 2.92 metre long and
0.495 metre wide rectilinear reaches connected at right angle by a
0.495 x 0.495 m square element. The channel slope is equal to zero. A
guillotine-type gate connects this L-shaped channel to a 2.44 x 2.39 m
Scientific Documentation
43
44
Figure 4.1
Table 4.1
Location
x (m)
y (m)
T1
1.19
1.20
T2
2.74
0.69
T3
4.24
0.69
T4
5.74
0.69
T5
6.56
1.51
T6
6.56
3.01
Validation
4.1.2
Numerical experiments
Simulations are performed using both the two-dimensional and the
three-dimensional shallow water equations.
An unstructured mesh is used containing 18311 triangular elements
and 9537 nodes. The minimum edge length is 0.01906 m and the
maximum edge length is 0.06125 m. In the 3D simulation 10 layers is
used for the vertical discretization. The time step is 0.002 s. At the
downstream boundary, a free outfall (absorbing) boundary condition is
applied. The wetting depth, flooding depth and drying depth are 0.002
m, 0.001 m and 0.0001 m, respectively.
A constant Manning coefficient of 105.26 m1/3/s is applied in the 2D
simulations, while a constant roughness height of 510-5 m is applied
in the 3D simulation.
4.1.3
Results
In Figure 4.2 time series of calculated surface elevations at the six
gauges locations are compared to the measurements. In Figure 4.3
contour plots of the surface elevations are shown at T = 1.6, 3.2 and
4.8 s (two-dimensional simulation).
In Figure 4.4 a vector plot and contour plots of the current speed at a
vertical profile along the centre line (from (x,y)=(5.7, 0.69) to
(x,y)=(6.4, 0.69)) at T = 6.4 s is shown.
Scientific Documentation
45
Figure 4.2
46
Time evolution of the water level at the six gauge locations. (blue) 3D
calculation, (black) 2D calculation and (red) Measurements by Frazo
and Zech (1999a,b)
Validation
Figure 4.3
Scientific Documentation
47
Figure 4.4
48
Vector plot and contour plots of the current speed at a vertical profile
along the centre line at T = 6.4 s
References
REFERENCES
Darwish M.S. and Moukalled F. (2003), TVD schemes for unstructured grids, Int. J. of
Heat and Mass Transfor, 46, 599-611)
Geernaert G.L. and Plant W.L (1990), Surface Waves and fluxes, Volume 1 Current
theory, Kluwer Academic Publishers, The Netherlands.
Hirsch, C. (1990). Numerical Computation of Internal and External Flows, Volume 2:
Computational Methods for Inviscid and Viscous Flows, Wiley.
Iqbal M. (1983). An Introduction to solar Radiation, Academic Press.
Jawahar P. and H. Kamath. (2000). A high-resolution procedure for Euler and NavierStokes computations on unstructured grids, Journal Comp. Physics, 164, 165-203.
Kantha and Clayson (2000). Small Scale Processes in Geophysical Fluid flows,
International Geophysics Series, Volume 67.
Munk, W., Anderson, E. (1948), Notes on the theory of the thermocline, Journal of Marine
Research, 7, 276-295.
Lind & Falkenmark (1972), Hydrology: en inledning till vattenressurslran,
Studentlitteratur (in Swedish).
Pugh, D.T. (1987), Tides, surges and mean sea-level: a handbook for engineers and
scientists. Wiley, Chichester, 472pp
Rodi, W. (1984), Turbulence models and their applications in hydraulics, IAHR, Delft, the
Netherlands.
Rodi, W. (1980), Turbulence Models and Their Application in Hydraulics - A State of the
Art Review, Special IAHR Publication.
Roe, P. L. (1981), Approximate Riemann solvers, parameter vectors, and differenceschemes, Journal of Computational Physics, 43, 357-372.
Sahlberg J. (1984). A hydrodynamic model for heat contents calculations on lakes at the ice
formation date, Document D4: 1984, Swedish council for Building Research.
Shu C.W. (1997), Essentially Non-Oscillatory and Weighted Essenetially Non-Oscillatory
Schemes for Hyperbolic Conservation Laws, NASA/CR-97-206253, ICASE
Report No. 97-65, NASA Langley Research Center, pp. 83.
Smagorinsky (1963), J. General Circulation Experiment with the Primitive Equations,
Monthly Weather Review, 91, No. 3, pp 99-164.
Scientific Documentation
49
Sleigh, P.A., Gaskell, P.H., Bersins, M. and Wright, N.G. (1998), An unstructured finitevolume algorithm for predicting flow in rivers and estuaries, Computers & Fluids,
Vol. 27, No. 4, 479-508.
Soares Frazo, S. and Zech, Y. (2002), Dam-break in channel with 90 bend, Journal of
Hydraulic Engineering, ASCE, 2002, 128, No. 11, 956-968.
Soares Frazo, S. and Zech, Y. (1999a), Effects of a sharp bend on dam-break flow, Proc.,
28th IAHR Congress, Graz, Austria, Technical Univ. Graz, Graz, Austria (CDRom).
Soares Frazo, S. and Zech, Y. (1999b), Dam-break flow through sharp bends Physical
model and 2D Boltzmann model validation, Proc., CADAM Meeting Wallingford,
U.K., 2-3 March 1998, European Commission, Brussels, Belgium, 151-169.
UNESCO (1981), The practical salinity scale 1978 and the international equation of state
of seawater 1980, UNESCO technical papers in marine science, 36, 1981.
Wu, Jin (1994), The sea surface is aerodynamically rough even under light winds,
Boundary layer Meteorology, 69, 149-158.
Wu, Jin (1980), Wind-stress Coefficients over sea surface and near neutral conditions A
revisit, Journal of Physical. Oceanography, 10, 727-740.
Zhao, D.H., Shen, H.W., Tabios, G.Q., Tan, W.Y. and Lai, J.S. (1994), Finite-volume twodimensional unsteady-flow model for river basins, Journal of Hydraulic
Engineering, ASCE, 1994, 120, No. 7, 863-833.
50