1. FLOODPLAIN REPRESENTATION AND ACCURACY OF PROCESSES 1D/2D
SESSION 2017
 FLOODPLAIN REPRESENTATION AND ACCURACY OF PROCESSES 1D/2D
Our question of the week was looking at the floodplain representation and accuracy of processes 1D/2D.
With the 2D model given as TELEMAC
Tessie BARJAT Anaïs MAYAU Tamsin JONES
Aymeric FRANCOIS Laura TALBA Konlavach MENGSUWAN
Andres GONZALEZ INIGUEZ Miquel SARRIAS MONTON
Oussama MOKHTARI Chenyu LIU
SUPERVISORS
TEAM MEMBERS
James CELRAY
Leslie SALVAN

2. INTRODUCTION Floodplain:
Area of low-lying ground adjacent to a river subject to flooding
Need to understand how water flows in these areas
Accuracy of processes:
Depends upon the model used
Ability to model ‘real processes’ occurring in the floodplain
Need to produce a 1D and 2D model – compare
3 November 2011 event chosen
In order to begin answering our question we began by defining the terms in our question
The floodplain is the area surrounding the river that is subject to flooding during high discharges
It is important to model the flow of water here as these areas of often exploited with buildings and agriculture
The accuracy of processes we defined as being dependent on the type of model used 1D or 2D
We must therefore produce an output of 1D and 2D models and compare their ability to model the floodplain
We chose the 2011 event as there was a large amount of data available and it is more recent then 1994, with no issues of gauge failure

3. INTRODUCTION
Aim:
Compare 1D and 2D model processes for the floodplain – find the limitations of processes represented.
Vs.
Objectives:
Discuss processes of 1D and 2D models
Produce an output for HEC-RAS (1D).
Produce an output for Telemac (2D).
Compare the representation of the floodplain for each model
In order to meet our aim we identified more main objectives
We split the group into 3 parts: 1 to discuss the theoretical processes of 1D and 2D models. Another group to produce outputs for HEC-RAS and another to produce outputs for TELEMAC
We would then work together to compare results between our models and to the theoretical findings
The var river

4. Mathematical governing equations
1D VS 2D MODELS
1D MODELS
2D MODELS
Cross-sections taken perpendicular to river flow (x)
Velocity is not considered to be uniform
Terrain represented as a continuous surface (x,y)
Uniform water level
across the cross-section
Mathematical governing equations
Both 1D and 2D models are based on mathematical governing equations
In 1D models many cross-sections are taken which describe the available volume in the system, the slope of channel bed and the roughness
1D models are different in that they are based on the assumption that flow is one-dimensional meaning that velocity is uniform across a cross-section
The water level is also assumed to be uniform for a given cross-section
With 2D models the terrain is represented as a continuous surface
Better for steep gradients
Uniform velocity across the cross-section
Water depth is not considered to be uniform
Better for low gradients

5. DOMAIN DEFINITION - HECRAS
Lower Reach of the Var River Basin → Covering the Floodplain
Upstream Boundary → La Manda Bridge (observed discharge data)
Downstream boundary → Mediterranean Sea

6. TOPOGRAPHY REPRESENTATION
Original DEM
DEM after bridges extraction

7. TOPOGRAPHY REPRESENTATION
WEIRS

8. CHOICE OF MODEL - HECRAS
Mike 11 case
Way to solve problem of uniform water level by the implementation of levees
Meters
Meters

9. HEC-RAS (MODEL SET-UP)
Topography
5m resolution DEM
ArcGIS
+HEC-GeoRAS
Cross sections have been obtained usig ArcGIS with HEC-GeoRAS toolkit. Sections have been drawn by hand keeping the perpendicularity and a similar space between them (about 100m)
Once created the file is imported on HEC-RAS
Cross sections

10. HEC-RAS (MODEL SET-UP)
Roughness coefficient
RIVER BED
n (s/m1/3)
Dense vegetation
0.080
Some vegetation
0.050
No vegetation
0.033
FLOOD PLAIN
Agricultural areas
Industrial areas
0.100
Urban areas
0.150
1- delimitate areas of similar roughness
2- assign a value to each area
3- overlay the cross ssections
4- assign each value to each part of the cross section

11. HEC-RAS (MODEL SET-UP)
Hydrograph
The hydrograph used as upstream boundary condition was observed on La Manda bridge on November 2011.

12. HEC-RAS (MODEL SET-UP)
Boundary Conditions
Upstream: Flow Hydrograph (La Manda bridge, 2011)
Downstream: Rating curve obtained from Manning equation, assuming a rectangular cross-section.
Initial Conditions
Initial distributed flow (20m3/s)

13. HEC-RAS (MODEL SET-UP)
Computation aspects
1D Courant–Friedrichs–Lewy condition:
∆x ≈ 100m
u ≈ 2m/s
Cmax = 1
∆t < 50s
17 days computed
30s computational time step
30min output interval
about 5 min to compute

14. HEC-RAS (FLOODPLAIN REPRESENTATION)
10 meters between cross sections
100 meters between cross sections

15. HEC-RAS (RESULT)

16. TELEMAC (MODEL SET-UP)
MANNING DISTRIBUTION
BUILDINGS DISTRIBUTION
Mesh characteristics:
25 m size
part of the sea
buildings
try to remove bridges elevation
Simulation parameters & equations:
2 boundary conditions
Numerical scheme Saint-Venant FE
Manning’s friction law
0.05
0.08
0.1
0.15
0.033

17. WATER DEPTH & VELOCITY AT THE BRIDGE
TELEMAC (RESULT)
WATER DEPTH & VELOCITY AT THE BRIDGE
Many observed problems for:
false flux observed
bridges impact the flood map
mesh size too big
time step too high
faced turbulence
TURBULENCE

18. TELEMAC (RESULT)
IMAGE
WATER DEPTHS CALCULATED
AIRPORT ZOOM

19. VELOCITY DISTRIBUTION
TELEMAC (RESULT)
IMAGE
VELOCITY DISTRIBUTION
AIRPORT ZOOM

20. CONCLUSION 1D: Quicker process time Limitations - uniform velocity
Accuracy - increase number of cross-sections (slower). Depends on modellers experience
2D:
Accounts for variability in the river
Limitations - computation time
Accuracy - finer mesh
Depends on the purpose - flood forecasting, scheme design
It is clear that 2D models are more accurate as they are able to account for variability in velocity and water level
Both models accuracy can be improved in 1D models we can increase the number of cross-sections, however this will increase computation time. Also have to think about the modellers experience as this will affect the ability to gain accurate cross-sections
In 2D models we can create a finer mesh
Our results suggest that Telemac is better at representing the floodplain in terms of accuracy of processes
However the main issue is that we do not know the purpose of the study. If the purpose is flood forecasting then we need rapid data so 1D may be better. If we want to design in the floodplain then we need accurate results so 2D may be better

21. We are grateful for the opportunity to take part in HydroEurope
QUESTIONS?
Thank you to all lecturers especially our supervisors - James and Leslie for their advice and experience
We are grateful for the opportunity to take part in HydroEurope