Presentation by Roy van Weerdenburg, Royal HaskoningDHV, Netherlands, at the Delft3D - User Days (Day 1: Hydrodynamics), during Delft Software Days - Edition 2017. Monday, 30 October 2017, Delft.

1. Case study of applying Delft3D FM in the
Waal River, the Netherlands
Delft3D User Days 2017
Roy van Weerdenburg
30 October 2017

2. Objectives of this study
 Analysis of the differences between D-Flow FM and Delft3D-FLOW
 Analysis of the modelling results of D-Flow FM compared to Delft3D-FLOW
 Hydrodynamics
 Morphology
 Discover new possibilities in D-Flow FM
 Get experience with the new modelling suite

3. Most important differences
 Flexible mesh: no rows and columns in the grid anymore ((m,n) coordinates)
 Allows for local refinement or coarsening of the grid
 ADI-solver no longer applicable in D-Flow FM
 Dynamic time step reduction based on CFL-condition
 All geometric model input in model-independent coordinates
 Independent of the grid
 DeltaShell is the GUI of D-Flow FM

4. Model-independent coordinates
 Grid enclosure is replaced by defining dry areas
 Currently no opportunity to see how exactly the input files are projected onto
the grid
D-Flow FM:
Spatial coordinates (*.xyz)
Delft3D-FLOW:
Directly projected on the grid

5. Validation models
 Deltares made many validation models to assess the performance of several
functionalities
 These models are very simplified situations
 e.g. a rectangular gutter or one single bend
 Investigate what the combined effect is in a more
complex model of a lowland river

6. Model set-up
 Delft3D FM Suite 2017 HMWQ (1.3.2.37884): β-version
 Set-up similar models in Delft3D-FLOW and D-Flow FM
 The new models are based on the DVR-model of the Rhine branches
 The same curvilinear grid was used in both models
 Input files were generated by the newest available version of Baseline

7. Hydrodynamic modelling
 Constant discharge: Q = 1600 m3/s (Lobith 2350 m3/s)

8. Hydrodynamic modelling
 Difference in stationary water level at the observation points (Q = 1600 m3/s)

9. Hydrodynamic modelling
 These differences are caused by
 The information in input files from Baseline
 Differences in the numerical solver
 Matlab toolbox to convert a Delft3D-FLOW model into D-Flow FM input files

10. Hydrodynamic modelling
 dflowfmConverter.m

11. Hydrodynamic modelling
 Difference in stationary water level at the observation points (Q = 1600 m3/s)
 Converted input files

13. Hydrodynamic modelling: first conclusions
 Differences in water level are partially caused by differing input files from
Baseline
 The new numerical solver calculates different water levels and the maximum
difference is approximately 5 cm during average discharge

14. Hydrodynamic modelling
 Constant discharge: Q = 820 m3/s (Lobith 1020 m3/s)
 This discharge is exceeded 95% of time
 Flow limited to the conveyance channel
 No influence of weirs and flood plains

15. Hydrodynamic modelling
 Difference in stationary water level at the observation points (Q = 820 m3/s)

16. Hydrodynamic modelling
 Difference in water level is now approximately 4.5 cm
 Relates to the reduction in water depth
 We consider the water levels and the flow velocities in four cross sections

17. Hydrodynamic modelling
KM 880
KM 885KM 902KM 904

18. left bank right bank

20. Hydrodynamic modelling
 The differences in flow velocities also occur at the downstream end

21. Hydrodynamic modelling
 Flow velocity vectors: blue = Delft3D-FLOW red = D-Flow FM

22. Hydrodynamic modelling: some conclusions
 The water level profiles in the cross sections look very similar
 Near the edges of the conveyance channel we see some differences in the
flow velocities

23. Hydrodynamic modelling: secondary flow
 The secondary flow pattern is in both models based on the primary flow, using
the same equations

24. Morphology
 Hydrodynamics directly affects the morphologic development
 Same numerical procedure for sediment transport rates in both models
 Upwind approach (default option)

25. Conclusions
 The input files as generated by Baseline cause water level differences of a
few centimeters
 The new numerical solving method computes water depths that differ up to 5
centimeters compared to the ADI-solver (average discharge)
 The new numerical solving method computes a slightly different pattern of
flow velocities in the cross sections
 The differences in morphological development are directly linked to the
hydrodynamics, because the equations and the solving method are the same

26. Conclusions
 The DeltaShell GUI provides much insight into the model components and
input may be changed easily within the interface
 Next step: Use field measurements to calibrate D-Flow FM model

27. Any further questions?
Roy van Weerdenburg
Student Hydraulic Engineering at Delft University of Technology
Intern at Royal Haskoning DHV
email: roy.van.weerdenburg@rhdhv.com
r.j.a.vanweerdenburg@student.tudelft.nl
phone: +31 6 46505353
Please contact me!