Transcript Henniger.ppt

Direct Numerical Simulation
of Particle Settling in Model Estuaries
R. Henniger(1), L. Kleiser(1), E. Meiburg(2)
(1) Institute
(2) Department
of Fluid Dynamics, ETH Zurich
of Mechanical Engineering, UCSB
Outline
 Introduction / motivation
 Computational setup

flow configuration
 governing equations and physical parameters
 simulation code
 Results


freshwater / saltwater mixing
particle settling
 Conclusions and outlook
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
2
Introduction
 Estuary mouth

light fresh-water
 heavy salt-water
 Suspended particles

e.g. sediment or pollutants
 transport out to the ocean
 particles settle and deposit
 Other influences

temperature profile
 Coriolis effect, tide, …
 Focus of the present study:
basic investigation of

freshwater / saltwater mixing
 particle transport, particle
settling and particle deposition
05/05/2009
Magdalena River (Colombia)
R. Henniger, L. Kleiser, E. Meiburg
3
Freshwater / saltwater mixing



Typically hypopycnal inflow
(Super-)critical?
Convective mixing, enhanced by
salty

turbulence in river
 Kelvin-Helmholtz or Holmboe waves
salt wedge
freshwater
salty
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
4
Particle load
freshwater + particles

hypopycnal:
salty
freshwater + particles

hyperpycnal:
salty
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
5
Particle transport
(1) Surface plume
(2) (Enhanced) particle settling

flocculation?
 turbulence enhanced settling?
(3) Bottom propagating turbidity
current
(1)
(2)
(3)
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
6
Model estuary configuration
salt sponge
convective outflow
inflow
symmetry planes
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
9
Governing equations, non-dimensional
 Incompressible Navier-Stokes and concentration transport equations
(in Boussinesq regime)
 Reynolds number:
2
32 3 2 3
H 1 0 0 G1
u1
b1
6
76 7 6 7
·
¸· ¸ ·
6 0 H 2 0 G 27 6 u27 6 b27
H G u
f
6
76 7 6 7
¢ =
6 0 0 H G 7 6 u 7= 6 b 7 ( )
D 0 p
0¸
3
3 5¢
4
4 35 4 35
D1 D2 D3 0
p
0
 Schmidt number:
 Richardson number:
 Particle settling velocity:
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
10
Physical parameters
reality
laboratory
simulation
Re
105-107
103-104
1500
Scsal
500-3000
500-3000
1
 turbulence
 “sharpness” of interfaces
Scpart
> Scsal
> Scsal
2
Risal
0.5-1
0.5-1
0.5
 sub-/supercritical flow
Ripart
< 0.05
< 0.05
0.05
 particle load
-us/U
< 10-2
< 10-2
0.01-0.02
turbulence
05/05/2009
interfaces
inertial forces
R. Henniger, L. Kleiser, E. Meiburg
 particle plume extent
loading
extent
11
Newly developed simulation code (summary)
 Incompressible flows + active scalars
 Discretization

compact finite differences in space
 explicit or semi-implicit time integration
 Massively parallel platform

3D domain decomposition (>95% parallel efficiency)
 sustained 16% peak performance on Cray XT
 scalability tested to up to 8000 cores and 17 billion grid points
 Validation

convergence orders in time and space
 convergence properties of iterative solvers
 temporal and spatial growth of eigenmodes
- channel flow
- shear layer flows with passive scalar

transitional and turbulent channel flow (vs. P. Schlatter)
 particle-driven gravity current (vs. F. Necker)
 parallel scaling properties
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
12
Results:
freshwater / saltwater mixing
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
13
Freshwater current
salt sponge
internal waves
Kelvin-Helmholtz
waves
csal = 0.75
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
14
Group velocity of internal waves
(measured with potential energy at y = 0)
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
15
Streamlines on water surface
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
16
Sub-/supercritical flow
 kinetic vs. buoyant forces
 measured with bulk Richardson number
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
17
Interface stability
 shear stress vs. density difference
 measured with gradient Richardson number
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
18
Results:
particle settling
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
19
Particle settling
 Three different settling velocities
us/U
= -0.02, -0.015, -0.01
 Qualitative agreement with laboratory experiments?

Maxworthy (JFM, 1999)
 Parsons et al. (Sedimentology, 2001)
 McCool & Parsons (Cont. Shelf Res., 2004)
 Open questions

extent of particle plume?
 particle settling modes (transient, steady state)?
 effective settling velocity?
 deposit profile?
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
20
Particle plume us/U
= -0.02,
cpart
= 0.1
x1
x1
x2
t = 300
t = 400
x1
x1
x2
t = 450
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
t = 600
21
Convective particle settling us/U
= -0.02
x1 = 26
x2 = 5
t = 300
x3
t = 350
x1
x2
x3
t = 400
x1
x2
x3
t = 600
x1
x2
x3
x1
05/05/2009
x2
R. Henniger, L. Kleiser, E. Meiburg
22
Particle mass
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
23
Effective particle settling velocity
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
25
Particle Deposit
us/U
05/05/2009
= -0.020, t = 710
R. Henniger, L. Kleiser, E. Meiburg
26
Particle Deposit
us/U
05/05/2009
= -0.015, t = 920
R. Henniger, L. Kleiser, E. Meiburg
27
Particle Deposit
us/U
05/05/2009
= -0.010, t = 1040
R. Henniger, L. Kleiser, E. Meiburg
28
Conclusions
 Definition of simulation setup

parameters
 inflow
 boundary conditions
 sponge zones, etc.
 Results: Basic effects compare well with laboratory experiments



freshwater-brine mixing
finger convection
enhanced convective particle settling
 Results obtained at moderate Re and Sc, accessible to DNS
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
29
Outlook
 Further increase of Re and Sc with LES in the future
 Implemented LES models:

ADM-RT model (filter model)
 (HPF) Smagorinsky
 (upwinding)
 Validation of LES to be completed
 Further option: more complex domains e.g. by

orthogonal curvilinear grids
 immersed boundary method
 (immersed interface method)
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
30
Appendix
05/05/2009
R. Henniger, L. Kleiser, E. Meiburg
31