1. Tuesday, November 4, 2014
NEXT GENERATION MODELLING APPLIED TO THE VENICE LAGOON
2nd BATCH PILOT CASE
G.T. Menel Lemos, E. Jeradi
Next Generation Hydro Software Symposium – 2014
2. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon
VENICE LAGOON: SENSITIVE ENVIRONMENT
Venice Town
Surface of 550 km2
8% of land
11% of channels
80% of mud flats, tidal shallows and salt marshes
Lido Inlet
Malamocco Inlet
Chioggia Inlet
3. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 1ST BATCH PILOT CASE
TASK 1: Delft3D-FLOW D-Flow FM*: CONVERSION
* D-Flow FM = hydrodynamic simulation engine of Delft3D Flexible Mesh
4. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 1ST BATCH PILOT CASE
TASK 2: Delft3D-FLOW D-Flow FM: WATER LEVELS
5. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 1ST BATCH PILOT CASE
TASK 2: Delft3D-FLOW D-Flow FM: FLUXES
6. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon – 2nd BATCH PILOT CASE
1.Define curvilinear grid for all the main channels
D-Flow FM VENICE MODEL: MESH CREATION
7. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
1.Define curvilinear grid for all the main channels
2.and sea area (with different resolutions)
D-Flow FM VENICE MODEL: MESH CREATION
8. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
1.Define curvilinear grid for all the main channels
2.and sea area (with different resolutions)
3.Define mesh of marshlands (based on samples)
D-Flow FM VENICE MODEL: MESH CREATION
9. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
D-Flow FM VENICE MODEL: MESH CREATION
1.Define curvilinear grid for all the main channels
2.and sea area (with different resolutions)
3.Define net of marshlands (based on samples)
4.Define net in the flat areas
10. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon – 2nd BATCH PILOT CASE
D-Flow FM VENICE MODEL: ORTHOGONALITY
11. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
~ 170.000 nodes 400.000 links
D-Flow FM VENICE MODEL: COMPLETE MESH
12. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
D-Flow FM VENICE LAGOON: NORTHERN PART
13. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE MODEL: BATHYMETRY
14. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE MODEL: BATHYMETRY
15. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
D-FLOW VENICE MODEL: NORTHERN PART
NEW D-Flow FM VENICE LAGOON MODEL: GRID OPTIMIZATION
16. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE TOWN MODEL: 1D CHANNELS
17. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE TOWN MODEL: 1D CHANNELS
18. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
COMPARISON BETWEEN MODELS AND MEASUREMENTS
Fluxes
Water levels
19. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
Node velocity
20. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
21. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
22. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE MODEL: RESULTS
Salinity
23. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
NEW D-Flow FM VENICE TOWN MODEL: RESULTS
Salinity
24. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
PERFORMANCE
CPUtime steps / Nr of TimeSteps
TtimeStep D-Flow FM/ TtimeStep Delft3D-FLOW
Run time D-Flow FM/Run time Delft3D-FLOW
Delft3D-FLOW
5 domains
D-Flow FM
optimized
Timestep (s)
3
auto (CFL)
Simulation period (s)
604800
604800
Nr of timesteps ( )
201600
223262
CPUtime steps (s)
89890.17**
88787.97
Run time (hh:mm:ss)
22:46:14
25:32:41
TtimeStep
0.44588
0.39769
SpeedFactor TimeStep
0.89190
SpeedFactor Wall Clock
1.12183
**Average cpu time of the 5 domains (not considering ddmodule wait time)
25. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon – 2nd BATCH PILOT CASE
D-Flow FM VENICE MODEL: MESH CREATION ORIGINAL SIN
26. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon – 2nd BATCH PILOT CASE
D-Flow FM VENICE MODEL: MESH CREATION DIFFICULTIES
Connection curvilinear to flat areas (very useful “External triangles to outer quads”)
Flat area –triangulation of samples to net makes… triangles!
Marshlands – a lot of samples so many triangles…
ORTHOGONALITY!!!
27. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ
Coupling procedure
Using offline coupling (communication file) and ddbound option to account for dd internal boundaries Pros:
•hydro run can have different duration and save only last cycle to communication file
•It is possible to aggregate the grid before coupling it Cons:
•Communication file can be very big
•Coupling with ddmodel can be tricky (need manual modifications of the *.hyd files)
Using online coupling by activating WaqInterval option in the mdu file Pros:
•No communication file: creation of WAQ files (.are, .vol, .flo, etc.) directly
•Simple procedure Cons:
•hydro run must have the same duration of the cycle for WAQ
•It is not possible to aggregate the net ** (**for the moment but it will be possible to create the WAQ files on an aggregated grid)
Delft3D-FLOW DD model
D-Flow FM
28. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ: TEST RUN
Hydrodynamic conditions:
•Average tide : semidiurnal +- 35 cm
•No wind
•Average river flows
WAQ conditions: processes initial sea/river bnd
•Continuity (g/m3) 1 1/1
•Salinity (g/kg) 34 35/0
•Cons. tracer (g/m3) 1 1/0
•Duration: 30 days,
•Dispersion: 1 m2/s
•Numerical scheme: 15 Iterative solver, backward diff.
1 min timestep
30 s timestep
29. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ: FLOCHECK
AFACT
BAGGER
Show per time step the multiplication factor
Applied to the cross-sectional area of each segment
Show the overall volume change
After the coupling procedure for the Delft3D-FLOW model a utility called “flocheck” has been used.
This utility makes a kind of “dredging” to increase the volume and/or lateral surface of isolate
segments in order to keep the residence time below the user-defined integration time-step.
1 min timestep
1 m2/s dispersion
30. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ: TEST RUN RESULTS – CONSERVATIVE TRACER
Delft3D-FLOW
D-Flow FM
Rivers are defined in different way in the two hydro models and so different in the WAQ:
•Delft3D-FLOW - as discharges
•D-Flow FM - discharges boundaries Same dispersion coefficient: 1 m2/s
31. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ: TEST RUN RESULTS - SALINITY
Delft3D-FLOW
D-FLOW
Rivers are defined in different way in the two hydro models and so different in the WAQ:
•Delft3D-FLOW - as discharges
•D-Flow FM- discharges boundaries Same dispersion coefficient: 1 m2/s
D-Flow FM
32. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon - 2nd BATCH PILOT CASE
DELWAQ: TEST RUN RESULTS - CONTINUITY
Delft3D-FLOW
D-Flow FM
1 min time step
30 s time step
33. Tuesday, November 4, 2014
Next Generation Modelling applied to the Venice Lagoon
COMMENTS
D-Flow FM:
•There was no easy way to de-refine easily the mesh. The existing method involves making a polygon and merging nodes and than smooth/orthogonalize again. The procedure is very time consuming in such complex system
•In conclusion the creation of the net should be well planned before starting the work DELWAQ:
•The online D-Flow FM - DELWAQ coupling procedure constrain to use the same grid for DELWAQ and Hydro simulations (and the same timestep…)
•It is not possible to use Dido to aggregate the mesh and the flocheck procedure doesn’t work
•It is not possible to defined an spatial varying initial condition using the D-Flow FM mesh
34. we plan the world of tomorrow
www.technital.it
THANK YOU!!
giselle.lemos@technital.it