FUNDAMENTAL EQUATIONS,

CONCEPTS AND IMPLEMENTATION

OF NUMERICAL SIMULATION IN
FREE SURFACE FLOW
Governing Equations of Fluid Flow
Navier-Stokes Equations
• A system of 4 nonlinear PDE of mixed
hyperbolic parabolic type describing the fluid
hydrodynamics in 3D.
• Three equations of conservation of momentum
in cartesian coordinate system plus equation of
continuity embodying the principal of
conservation of mass.
• Expression of F=ma for a fluid in a differential
volume.

2
 u u u u
ax  u v w 1
t x y z
v v v v
ay   u  v  w  2
t x y z
w w w w
ax  u v w  3
t x y z
ui ui
ai  uj
t x j
3
• The acceleration vector contains local
acceleration and covective terms
• The force vector is broken into a
surface force and a body force per
unit volume.
• The body force vector is due only to
gravity while the pressure forces and
the viscous shear stresses make up
the surface forces.

4
 1  p  xx  yx  zx 
fx  gx       ( 4)
  x x y z 
1  p  xy  yy  zy 
f y  g y       (5)
  y x y z 
1  p  xz  yz  zz 
f z  g z       (6)
  z x y z 

5
• The stresses are related to fluid element
displacements by invoking the Stokes viscosity
law for an incompressible fluid.
u v w
 xx  2  ,  yy  2  ,  zz  2   7
x x x
 u v 
 xy   yx       8
 y x 
 w u 
 xz   zx       9
 x z 
 v w 
 yz   zy      10
 z y  6
Substituting eqs. 7-10 into eqs. 4-6, we get

1 p   2u  2u  2u 
fx  gx     2  2  2  (11)
 x  x y z 
1 p   2v  2v  2v 
fy  gy    2  2  2  (12)
 y  x y z 
1 p  2w 2w 2w 
fz  gz    2  2  2  (13)
 z  x y z 
1 p  2 ui
fi  gi   Einstein notation
 xi x j x j
7
 The three N-S momentum equations can
be written in compact form as
ui ui  1 p  2ui
uj    g i (A1 )
t x j  xi x j x j

The equation of continuity for an incompressible

fluid

u v w
  0
x y z
ui
0 (A 2)
xi 8
 Turbulence
The free surface flows occurring in nature is almost
always turbulent. Turbulence is characterized by
random fluctuating motion of the fluid masses in
three dimensions. A few characteristic of the
turbulence are:
1. Irregularity
Turbulent flow is irregular, random and chaotic. The
flow consists of a spectrum of different scales (eddy
sizes) where largest eddies are of the order of the
flow geometry (i.e. flow depth, jet width, etc). At the

9
 other end of the spectra we have the smallest
eddies which are by viscous forces (stresses)
dissipated into internal energy.
2. Diffusuvity The turbulence increases the
exchange of momentum in flow thereby
increasing the resistance (wall friction) in internal
flows such as in channels and pipes.
3. Large Reynolds Number Turbulent flow occurs
at high Reynolds number. For example, the
transition to turbulent flow in pipes occurs at
NR~2300 and in boundary layers at NR~100000

10
4.Three-dimensional Turbulent flow is always three-
dimensional. However, when the equations are
time averaged we can treat the flow as two-
dimensional.
5. Dissipation Turbulent flow is dissipative, which
means that kinetic energy in the small (dissipative)
eddies are transformed into internal energy. The
small eddies receive the kinetic energy from
slightly larger eddies. The slightly larger eddies
receive their energy from even larger eddies and
so on. The largest eddies extract their energy from
the mean flow. This process of transferred energy
from the largest turbulent scales (eddies) to the
smallest is called cascade process.
11
 Turbulence
. The random , chaotic nature of turbulence is
u  u  u
treated by dividing the instantaneous
v  v  v
values of velocity components and
pressure into a mean value and a w  w  w
fluctuating value, i.e. p  p  p
Why decompose variables ?
Firstly, we are usually interested in the mean values
rather than the time histories. Secondly, when we want
to solve the Navier-Stokes equation numerically it would
require a very fine grid to resolve all turbulent scales
and it would also require a fine resolution in time since
turbulent flow is always unsteady.
12
 Reynolds Time-averaged Navier-Stokes
Equations

These are obtained from the N-S equations and

include the flow turbulence effect as well.

u u u u 1 p   u 
u v w  gx     uu   
t x y z  x x  x 
  u    u 
  uv       uw   
y  y  z  z 
Momentum equation in x  direction
13
 RNS Equations

u i u i 1  p  vij 1  Rij
uj     gi
t x j  xi x j  x j
u i
0
x
 u i u j 
where  vij      viscous stress tensor
 x x 
 j i 

 Rij    uiu  j Reynold stress tensor

14
 Reynold Stresses
The continuity equation remains unchanged except
that instantaneous velocity components are
replaced by the time-averaged ones. The three
momentum equations on the LHS are changed only
to the extent that the inertial and convective
acceleration terms are now expressed in terms of
time averaged velocity components. The most
significant change is that on the LHS we now have
the Reynold stresses. These are time-averaged
products of fluctuating velocity components and are
responsible for considerable momentum exchange
in turbulent flow.

15
 Closure Problem
• 3 velocity components, one pressure and
6 Reynold stress terms = 10 unknowns
• No. of equations=4
• As No. of unknowns >No. of equations,
the problem is indeterminate. One need to
close the problem to obtain a solution.
• The turbulence modeling tries to represent
the Reynold stresses in terms of the time-
averaged velocity components.
16
 Turbulence Models
Boussinesq Model
An algebraic equation is used to compute a
turbulent viscosity, often called eddy viscosity.
The Reynolds stress tensor is expressed in
terms of the time-averaged velocity gradients
and the turbulent viscosity.

 U j U i 
 uiu j   t   
 xi x 
 j 

17
 k-ε Turbulence Model
Two transport equations are solved which describe
the transport of the turbulent kinetic energy, k and
its dissipation, ε. The eddy viscosity is calculated as

k
 t  c 2

the Reynold stress tensor is calculated
via the Boussinesq approximation

18
 RNS Equations and River Flow
Simulation
• RNS equations are seldom used for the
river flow simulation. Reasons being
• High Cost
• Long Calculation time
• Flow structure
• Method of choice for flows in rivers,
streams and overland flow is 2D and 1D
Saint Venant equations or Shallow water
equations
19
 2D Saint Venant Equation

• Obtained from RNS equations by depth-

averaging.
• Suitable for flow over a dyke, through the
breach, over the floodplain.
• Assumptions: hydrostatic pressure
distribution, small channel slope,

20
2D Saint Venant Equations
h  (hu )  (hv)
   0 continuity eq.
t x y
 (hu )  (hu 2 )  (huv) h zb
   gh   gh
t x y x x
u 2  v2
 gn 2u 1
x momentum eq
h 3
 (hv)  (huv)  (hv )
2
h zb
   gh   gh
t x y y y
u v
2 2
 gn v
2
1
y momentum eq 21
h 3
1D Saint Venant Equation
Q A
 0
t t
U U h
U g  g (S0  S f )
t x x
The friction slope Sf is usually
obtained from a uniform flow formula
such as Manning or chezy.

22
 Simplified Equations of Saint
Venant

1 u u u h
   So  S f
g t g x x
dynamic dyn. quasi diffusive kinematic
wave steady wave wave wave

23
 Relative Weight of Each Term
in SV Equation
1 u 5
Order of magnitude of O(10 )
each term In SV equation g t
for a flood on river Rhone u u
O(10 5 )
g x
So O(10 3 )
Sf O(10 3 )

24
 Calculation Grid
• Breaking up of the flow domain into small
cells is central to CFD. Grid or mesh refers
to the totality of such cells.
• In Open channel flow simulation the
vertices of a cell define a unique point
(x,y,,z)
* The governing equations are discretized
into algebraic equations and solved over
the volume of a cell.
25
 Classification of Grids
• Shape
• Orthogonality
• Structure
• Blocks
• Position of variables
• Grid movements

26
 Boundary Conditions
• Inflow b. c
If Fr<1, specify discharge or velocity.
If Fr>1, specify discharge or velocity and
depth
• Outflow b.c
Zero depth gradient or Newmann b.c
Specify depth
Specify Fr=1
27
 Initial Condition
• Values of flow depth, velocity, pressure etc
must be assigned at the start of the
calculation run.
• Hot start

28
 Wall Boundary Condition
No slip condition
u 1  30 y 
require very fine  ln  
meshing adjacent to u 
  ks 
the wall requiring lot wall law for rough boundaries
of CPU time. Flow
close to the wall is not
resolved but wall laws
derived from the
universal velocity
distribution are used.
29
 Methods of Solution
• Finite Difference
Method
• Finite element method
• Finite volume method
• Strategies
• Implicit
• Explicit
• CFL condition
dx
dt 
(U  c) 30