The application of a two-dimensional sediment transport

model in a Cumberland Plateau mountainous stream reach

with complex morphology and coarse substrate
Those who thoughtlessly make use of the miracles of science and
technology, without understanding more about them than a cow eating
plants understands about botany, should be ashamed of themselves.
(Einstein).
Daniel Johnson and Dr. John Schwartz
SRI conference
November, 5
th
2008
Coastal Science & Engineering
Background
USACOE
A calibrated computational model can
facilitate the design process.
A computational model can be a powerful
tool for investigating sediment transport
and in-stream hydrodynamic processes
(Chen et al., 2007; Duan et al., 2007; Jia
et al., 1999; Jia et al., 2006; Langendoen,
2001; Papanicolaou et al., 2008; Scott et
al., 2005; White, 2008; Wu, 2008).
Two-dimensional sediment transport modeling
If the vertical variations of flow and sediment quantities in a water body are sufficiently small or
can be determined analytically, their variations in the horizontal plane can be approximately
described by a depth-averaged 2D model (Wu, 2008).
CCHE2D Output
Abbreviated (2D)
Simulate the transverse and longitudinal
velocity component
Structured grids
Increased complexity necessitates
significantly more data input
Does not require the computational
capacity of a 3D model
2D depth averaged models assume the
vertical variations of flow and sediment
quantities are sufficiently small.
Incorporates turbulence through depth
averaged eddy viscosity coefficients
Examples: CCHE2D, DELFT2D,
FLUVIAL 12, and MIKE 21
Study background
Study Objective
Determine the sensitivity of model output (at various scales) to the Mannings n variable and
upstreamsediment rating curves.
Scales of interest:
Reach scale > 100 m
Local scale ~ 10 m
Point scale ~ 1 m
Ligias Creek (study site)
CCHE2D (model)
Study site (Ligias Creek)
2
1
2
1
2
CCHE2D
Two-dimensional, depth averaged, with unsteady flow and sediment transport capabilities.
Developed at the National Center for Computational Hydroscience and Engineering (NCCHE)
includes a Graphical Users Interface (CCHE-GUI) and a structured mesh generator (CCHE2D Mesh
Generator).
Simulates both suspended load and bedload transport
Solves the Navier-Stokes equations
Input data requirements
What is required to perform a sediment transport
simulation?
Channel geometry (physical domain described
by a structured mesh)
Upstream and downstream hydraulic boundary
conditions
Calculated Mannings n value
Discharge or velocity measurements
Measured cross sectional area
Measured wetted perimeter
Measured Energy slope
Upstream sediment boundary conditions
Bedload rating curve
Suspended load rating curve
2 / 1 3 / 2
S R
n
k
V
n
Mannings equation
(Langendoen, 2000)
Channel geometry & monitoring points
Developed mesh (3D view) [looking upstream from downstream
modeling boundary]
In-stream velocity & energy slope measurements
Marsh-McBirney Flo-mate Model
2000 Flowmeter.
Ligias Fork
Stage-Discharge Plot
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
2.75
0 10 20 30 40 50 60 70 80 90 100
Stream Discharge (m
3
/sec)
S
t
r
e
a
m

S
t
a
g
e

R
e
a
d
i
n
g

(
m
e
t
e
r
s
)
Measured
Simulated
Global Water Stage Recorders
(Model No. WL-16)
Energy Slope vs. Time
0.0038
0.0042
0.0046
0.0050
0.0054
0.0058
0.0062
0.0066
03/23 03/28 04/02 04/07 04/12 04/17 04/22 04/27 05/02 05/07 05/12
Time (mm/dd)
E
n
e
r
g
y

S
l
o
p
e

(
f
t
/
f
t
)
92.750
93.250
93.750
94.250
94.750
95.250
95.750
96.250
96.750
97.250
97.750
98.250
98.750
99.250
ENERGY SLOPE DS WSEL US WSEL
Measured water surface and calculated energy slope
Stream stage (m) versus discharge (cms)
Suspended load sampling & rating curve
H-48 Depth Integrated Sediment
Sampler (Model 5200)
Suspended sediment rating curve
Bedload trap (net-frame sampler)
1. Allow for the collection of a physical sample for sieve analysis
2. Can be utilized for a long sampling duration
3. Portable
4. Require minimal stream excavation
Bunte K., Abt S.R., Potyondy J.P., and S.E. Ryan. (2004). Measurement of Coarse Gravel and
Cobble Transport Using Portable Bedload Traps. Journal of Hydraulic Engineering, Vol. 130,
No.9, pp. 879-893
Net-frame sampler
Base
Aluminum trap
Trailing net
Bedload sample Bedload sample
Net-frame sampler in action
Calculated bedload rating curve
Meyer-Peter and Muller
Energy slope
Critical diameter
Water depth
Shields parameter
Critical value of shields parameter (0.047)
Hiding coefficient, equal to a value of 0.7, was implemented to account for the bimodal bed
gradation and the influence of hiding and exposure effects on sediment transport.
2
3
*
3
) 047 . 0 ( 0 . 8
) 1 (
s
b
b
gd SG
q
s
d SG
S y
1
0
*
Meyer-Peter and Muller empirical formula
(Sturm, 2001)
Shields parameter
(Sturm, 2001)
Summary of methods
Bed elevation change was measured at multiple
monitoring points after a significant storm event.
A bedload rating curve was calculated (Meyer-Peter
& Muller equation) and applied at the upstream
boundary
A suspended load rating curve was developed
based on measured suspended sediment samples
A mannings n value was calculated based on
available energy slope data
Initial simulations began with a Mannings n value
equal to 0.025 (value calculated based on
measured energy slope data) and calculated
sediment rating curves (100% bedload, 100%
suspended load).
Additional model simulations were performed by
varying the Mannings n variable and upstream
sediment rating curves.
Simulated and measured bed elevation change
results were compared at the reach, local, and point
scales.
Multivariate statistical methods were implemented
to determine the correlation and sensitivity of model
output (bed elevation change) to the various
independent variables (Mannings n variable,
suspended sediment rating curve, bedload rating
curve)
Description
Manning's n
value Description
Manning's n
value
0% suspended, 0% Bedload 0.01 10% suspended, 0% Bedload 0.025
0% suspended, 0% Bedload 0.015
0% suspended, 0% Bedload 0.02 10% suspended, 10% Bedload 0.055
0% suspended, 0% Bedload 0.025 10% suspended, 10% Bedload 0.065
0% suspended, 0% Bedload 0.03
0% suspended, 0% Bedload 0.035 100% suspended, 0% Bedload 0.020
0% suspended, 0% Bedload 0.040 100% suspended, 0% Bedload 0.025
0% suspended, 0% Bedload 0.041 100% suspended, 0% Bedload 0.035
0% suspended, 0% Bedload 0.042 100% suspended, 0% Bedload 0.045
0% suspended, 0% Bedload 0.043 100% suspended, 0% Bedload 0.055
0% suspended, 0% Bedload 0.044
0% suspended, 0% Bedload 0.045 20% suspended, 0% Bedload 0.025
0% suspended, 0% Bedload 0.050
0% suspended, 0% Bedload 0.055 25% Suspended, 5% Bedload 0.050
0% Suspended, 1% Bedload 0.025 30% suspended, 0% bedload 0.025
0% Suspended, 1% Bedload 0.035
40% suspended, 0% bedload 0.025
0% Suspended, 10% Bedload 0.025
5% suspended, 5% Bedload 0.025
0% Suspended, 20% Bedload 0.025 5% suspended, 5% Bedload 0.035
5% suspended, 5% Bedload 0.045
0% Suspended, 30% Bedload 0.025
50% suspended, 50% Bedload 0.010
0% Suspended, 40% Bedload 0.025 50% suspended, 50% Bedload 0.025
50% suspended, 50% Bedload 0.050
0% Suspended, 5% Bedload 0.025 50% suspended, 50% Bedload 0.055
0% Suspended, 5% Bedload 0.035
0% suspended, 5% Bedload 0.055
Manning's n value Bedload Suspended load
0.01 - 0.065 0% - 50% 0% - 100%
Successful model simulations (variable range)
Modeling results
Bed elevation change as a function of channel roughness
0% Suspended load, 5% Bedload, 0.025 Mannings n value
0% Suspended load, 5% Bedload, 0.055 Mannings n value 0% Suspended load, 0% Bedload, 0.05 Mannings n value
0% Suspended load, 0% Bedload, 0.015 Mannings n value
Modeling results
Bed elevation change as a function of sediment inflows
50% Suspended load, 50% Bedload, 0.025 Mannings n value
5% Suspended load, 5% Bedload, 0.025 Mannings n value
0% Suspended load, 0% Bedload, 0.025 Mannings n value
Study findings
Extensive effort and time are required to obtain
reasonable model input data
bedload sampling is particularly problematic
Critical field measurements include:
time varying bedload transport rates as a
function of discharge
energy slope as a function of discharge
(Mannings n calculation)
and discharge as a function of time
It is critical to obtain sufficient measurements
during peak discharges or seasonally high
discharge rates
The model appears most sensitive to the bedload
sediment rating curve at the upstreamboundary.
If the bedload rating curve is not reasonable,
sediment mass builds at the entrance to the
finite element cells producing increased bed
elevations, a hydraulic backwater at the
upstream boundary, and ultimately numerical
instability.
The model appears most sensitive to the
Mannings n value at the downstream modeling
boundary.
When the roughness value is small, depth
averaged local velocities increase and erosion
accelerates near the downstream boundary.
Sediment mass accumulating at the upstream boundary, hydraulic
backwater and instability
Erosion accelerating near the downstream
boundary
2D modeling & stream restoration design
Reliance on 2D sediment models without adequate
data collection and model calibration may lead to
poorly designed stream restoration projects, and
ultimate project failure.
In complex channels, 2D model performance at the
local, reach, or point scales is likely not possible with
current technology. However, in less complex
channels others have found existing 2D models work
reasonably well.
It appears the current state-of-practice for stream
restoration must rely on 1D reach-scale sediment
transport models, with a general goal to determine
whether a reach is sediment-supply or capacity
limited.
2D models may be best implemented to analyze local
scale hydrodynamic processes.
sediment transport can then be analyzed
analytically based on the hydraulic results
produced by the 2D model.
In summary, it is very hard for a computational model
to precisely mimic the chaos and complex sediment
transport processes which occur in a natural channel.
Two-dimensional simulation results
The application of a two-dimensional sediment transport
model in a Cumberland Plateau mountainous stream reach
with complex morphology and coarse substrate
Those who thoughtlessly make use of the miracles of science and
technology, without understanding more about them than a cow eating
plants understands about botany, should be ashamed of themselves.
(Einstein).
Daniel Johnson and Dr. John Schwartz
SRI conference
November, 5
th
2008
Coastal Science & Engineering