dsd-int 2014 - openmi symposium - a selection of water-related applications of openmi, bernhard...
DESCRIPTION
OpenMITRANSCRIPT
A selection of water-related applications of OpenMI Bernhard Becker ([email protected]) Geert Prinsen Jan Talsma Quanduo Gao Neeltje Goorden
OpenMI connects different domains
1. Open channel flow – mechanical engineering
2. Open channel flow – real-time control
3. Surface – subsurface water interaction
Becker, B.; Dahm, R.; van Heeringen, K.-J.; Goorden, N.; Kramer, N.; Kooij, K.; Gooijer, J.; Jansen, J. (2012): Op zoek naar een optimaal ontwerp voor een groot uitwateringsgemaal in het Lauwersmeer. H2O 44 (13), 11–13.
Becker, B. P.; Schüttrumpf, H. (2010): An OpenMI module for the groundwater flow simulation programme Feflow. Journal of Hydroinformatics 13 (1), 1–13.
Schwanenberg, D.; Becker, B. P. .; Schruff, T. (2011): SOBEK-Grobmodell des staugeregelten Oberrheins. Report No. 1201242-000-ZWS-0014. Deltares.
OpenMI bridges institutional gaps
1. Technical exercise: external and iterative coupling with OpenMI
2. Water authorities Fryslân and Noorderzijlvest
3. Coupling of multiple open channel models
4. Nesting: large scale and detail models
Becker, B.; Talsma, J. (2013): On the external and iterative coupling of multiple open channel flow models with OpenMI. Revista de Ingeniería Innova 6, 51–66.
Becker, B.; Gao, Q. (2012): Koppelen Sobek-modellen “Wetterskip Fryslân” en “Waterschap Noorderzijlvest” via OpenMI. Report No. 1204514-000-ZWS-0007. Deltares, Delft.
0100200300400
500600700800
2 sub-models 3 sub-models 4 sub-models 5 sub-models 6 sub-models
com
putin
g tim
e in
sec
onds
implicit (Sobek CMT)explicit (OpenMI) Becker, B. P. J.; Talsma, J.; Gao, Q.; Ruijgh, E. (2012): Coupling
of multiple channel flow models with OpenMI. in: Proceedings of 10th International Conference on Hydroinformatics; Hamburg, Germany.
Gao, Q.; Prinsen, G. F. (2014): Koppeling SOBEK-2 en SOBEK-3 voor het LSM via OpenMI. Deltares rapport No. 1209449-009-ZWS-0003. Deltares, Delft.
Open channel flow mechanical engineering
Design of a pump station for lake Lauwersmeer (the Netherlands)
- more extreme rainfall events and rising sea level expected
- drainage of polder areas must be facilitated with a pump station
Sobek
Open channel flow mechanical engineering
Water level (see and inland)
Power demand for pumps
Discharge sluices and pumps
Open channel flow real-time control
Control of the Oberrhein (upper Rhine) water system: Decision tree and open channel system
Sobek RTC-Tools Channel flow (Q, h) Water system state Control parameter Control parameter (crest level, turbine discharge)
Sobek: open channel flow RTC-Tools: human operations (control)
Open channel flow real-time control
Simulation period: 14 days
Inundation area groundwater Ilmoflood Feflow
head leakage flow flow
Two coupled simulations: Ilmoflood Feflow saturated Ilmoflood Feflow unsaturated
Surface – subsurface water interaction
OpenMI compliance of FEFLOW
Feflow (DHI Wasy GmbH): control via the interface manager and remote procedure calls
Becker & Schüttrumpf, JHydroInf, 2011
Challenge: Source code not available
Technical exercise: two Sobek models
47.048.049.050.051.052.053.054.055.056.057.0
03-01-010:00
05-01-010:00
07-01-010:00
09-01-010:00
11-01-010:00
13-01-010:00
15-01-010:00
17-01-010:00
19-01-010:00
time
Wat
er le
vel [
m a
d]
Waterlevel (m AD)
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
3500.0
31-12-000:00
05-01-010:00
10-01-010:00
15-01-010:00
20-01-010:00
25-01-010:00
30-01-010:00
time
Disc
harg
e [m
³/s]
Discharge (m³/s)
49.050.051.052.053.054.055.056.057.058.059.060.0
03-01-010:00
05-01-010:00
07-01-010:00
09-01-010:00
11-01-010:00
13-01-010:00
15-01-010:00
17-01-010:00
19-01-010:00
time
Wat
er le
vel [
m a
d]
Waterlevel (m AD)
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
3500.0
03-01-010:00
05-01-010:00
07-01-010:00
09-01-010:00
11-01-010:00
13-01-010:00
15-01-010:00
17-01-010:00
19-01-010:00
time
Disc
harg
e [m
³/s]
Discharge (m³/s)
sub-model “channel west” sub-model “channel east“
input discharge output discharge
output water level input water level
Exchange of results and boundary conditions between multiple models
Request-reply-mechanism
The model component that asks first computes last. The model that asks gives the guess. The quantity which is computed first has the same value in both of the models.
Sobek EastSobek West
6
3
2
7
8z
4
1 Trigger
5
solution
solution
requestreply
runtime
t = 0
t = 0
t = 1
t = 1
t = tend
t = tend
h?
Q?
Q!
h!
h?
External coupling (time-lagged coupling)
Position of the trigger impacts the result external coupling has a time lag
Iterative coupling
OpenMI compliant component “iteration controller”
Iterative coupling
Iterations increase accuracy by repeating data exchange per time step
10 november 2014
Iterative and external coupling: test case results
External coupling as accurate as iterative coupling, but more controllable and less computational expensive
coupling method (min) computing time ternally coupled (10) 30 s externally coupled (10) 27 s externally coupled (30) 14 s externally coupled (30) 12 s iteratively coupled (10) 192 s iteratively coupled (30) 50 s iteratively coupled (60) 43 s implicit solution 3 s
Water authorities Fryslân and Noorderzijlvest
Boezemmodellen Wetterskip Fryslân and Noorderzijlvest coupled at three connection points
One water system, two water authorities
Comparison at connection nodes
-200
20406080
100120140160180
21-10 23-10 25-10 27-10 29-10 31-10 02-11 04-11 06-11tijd
debi
et Q
[m³/s
]
020406080100120140160180
vers
chil
debi
et [m
³/s]
Q_WF_uncoupled Q_WF_coupled Q_NZV_uncoupledQ_NZV_coupled verschil_uncoupled verschil_coupled
-1.20-1.00-0.80-0.60-0.40-0.200.000.200.40
21-10 23-10 25-10 27-10 29-10 31-10 02-11 04-11 06-11tijd
wat
erst
and
h [m
]
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
vers
chil
wat
erst
and
[m]
h_WF_uncoupled h_WF_coupled h_NZV_uncoupledh_NZV_coupled verschil_uncoupled verschil_coupled
10 november 2014
Multiple open channel flow models
From 2 to 6 model coupled:
Multiple models: performance
0100200300400
500600700800
2 sub-models 3 sub-models 4 sub-models 5 sub-models 6 sub-models
com
putin
g tim
e in
sec
onds
implicit (Sobek CMT)explicit (OpenMI)
• Water level differences: less than 3 cm
• Discharge differences: less than 1 m3/s
• Computational effort of explicit coupling increases disproportionally with the number of sub-models
implementation of OpenMI standard into Sobek has high potential to increase performance (see next slide!)
Nesting: large-scale and detail models
National model Model Rijntakken
Model Maas
Nested models
Can we use OpenMI to compose a National model with the detail models?
- Models have different states of maintenance - Models have different grid refinement - Models use different versions of Sobek (Sobek 3 and Sobek 2)
Model areas overleap
Nested model: performance
Simulation time one year, time step 10 min: three models coupled via OpenMI: 33.0 hours one model: 30.5 hours Only 10 % more with OpenMI!
Conclusion
OpenMI connects different water domains: - Open channel flow - Mechanical engineering - Real-time control - Surface/subsurface water OpenMI facilitates the modelling of interaction processes
OpenMI connects model borders - Models are owned by different institutions - Models have different states of maintenance (including software
versions) - Models have different degree of grid refinement OpenMI gives flexibility on scales and maintenance