Embed Size (px)
Citation preview
Real-time river flood control
with Model Predictive Control (MPC)Implemented for the river Demer basin (Belgium)
Patrick Willems
KU Leuven
Flood of September 1998
Mitigation strategies
Local solutionsDykes, rectifying river stretches, …
1959: ir. Roovers “enhanced water flow”
Give space to the river1970’s: Schulensmeer
Regulated hydraulic infrastructure
Early warning systems2004: Operational basin model
2014: Launch waterinfo.be
Intelligent real-time control
Study case & Approach
OBM Demer
(InfoWorks RS)
Conceptual model
Model Predictive
Control
Optimal control
strategy
Observations
Rainfall
forecasts
3 PhDs on the topic
Toni Barjas-Blanco, 2010. ‘The Demer controlled by MPC’, KU Leuven – Faculty of
Engineering (prom. B. De Moor, J. Berlamont, P. Willems)
Maarten Breckpot, 2013. ‘Model predictive control of river flooding’, KU Leuven –
Faculty of Engineering (prom. B. De Moor, P. Willems)
Evert Vermuyten, 2018. ‘Real-time flood control by means of model predictive
control and a reduced genetic algorithm’, KU Leuven – Faculty of Engineering
(prom. P. Willems)
Operational implementation at this moment
By Flemish Environment Agency (VMM)
Approach developed by KU Leuven PhDs
Companies involved in current operational implementation:
Antea Group, IMDC, Fabricom, IPCOS
Conceptual models
• Integrated approachInclude hydrology, rivers, sewers
and other processes
• Super fast
• Similar accuracy as full HD
model if properly set up
and calibrated
Conceptual model: network
Hydrodynamic InfoWorks model (OBM) Conceptual model
• Data-based mechanistic approach
• Storage cell concept
• Modular setting
Conceptual model: different levels of aggregation
Conceptual model: different levels of aggregation
Conceptual model structures & parameters
Linear
reservoir
River / Floodplain
Sewer
Other• Flows
• Water levels
PWL
• Close water balance
Continuity
equation
• Flows
• Rainfall runoff
Transfer
function
• Flows
Static /
Dynamic
• Flows
• Water levels
• Rating curves
ANN
• Flows river -
floodplain
ANFIS
• Flows
Hydraulic
structures
• Gate regulations
PLC
• Water levels
• Rating curves
M5’ trees
• Rating curves
SDP
Conceptual model: results
Conceptual model: results
InfoWorks
RS
Conceptual
model
WL nodes > 3 500 466
Hydraulic
structures> 1 700 391
Reservoirs / 90
Computation
time1h15m 0.8s
Conceptual model: results
Conceptual model: results
Real time control by means of RGA-MPC
• Model Predictive Control (MPC)
o River model
o Flow & water level observations
o Rainfall forecast
o Optimizer
• Reduced Genetic Algorithm (RGA)
o Based on standard GA
o Objectives:
• Retention basin dikes
• Damage cost
• Critical dikes
• Retention basins
timet t+Δt
RIVER
MODEL
SCENARIO
GENERATOR
SELECTOR
rainfall
forecasts
new GL
scenario
observations
predicted
system states
best GL
scenario
optimal GL
scenario
Gate Level (GL) scenario generation
RGA versus GA
MPC results (perfect forecasts assumed)
Event
Economic damage
cost [€]Damage
reduction
[%]PLC MPC
Sept1998 3.0M 2.1M 30Aug2003 0 0 /Dec1999 0 0 /Jan1995 0 0 /Jan2002 0 0 /Nov2010 300 100 (67)
VMM 4.7M 3.5M 26T1000 2.0M 1.7M 15
Sept1998x1.3 28.0M 26.8M 42xSept1998 28.1M 27.5M 2
MPC results Sept 1998
MPC resultsRiver Herk subbasin:
MPC results: after climate change ?River Herk subbasin:
Uncertainties involved
• Hydrodynamic model uncertainty
o Model structure, calibration errors, seasonal vegetation, …
o Data Assimilation: State Estimators, Prediction Error Methods
• Input uncertainty
o Rainfall forecasts, hydrological model uncertainty, …
o Robust MPC methods: MMPC, AMMPC, TB-MPC, …
Influence rainfall forecast uncertainty
River Herk subbasin:
Influence rainfall forecast uncertainty
River Herk subbasin:
Conclusions
• Conceptual modelling
o Integrated approach
o Flexible model detail
o Super fast
o Accuracy similar to full hydrodynamic model
• RGA-MPC for real time flood control
o Outperforms PLC
o Computationally very efficient
o Can handle large and complex networks
o Uncertainties to be dealt with
Main references:
On conceptual modelling method:
Wolfs, V., Meert, P., Willems, P. (2015). Modular conceptual modelling approach and software for
river hydraulic simulations. Environmental Modelling and Software, 71, 60-77
On RGA-MPC approach:
Vermuyten E., Meert P., Wolfs V., Willems P. (2018). Combining model predictive control with a
reduced genetic algorithm for real-time flood control. Journal of Water Resources Planning and
Management, 144(2), doi:10.1061/(ASCE)WR.1943-5452.0000859 (in press)
Contact:
Other references:
Chiang, P., Willems, P. (2013), ‘Model conceptualization procedure for river (flood) hydraulic computations: Case study of
the Demer River, Belgium’, Water Resources Management, 27(12), 4277–4289
Chiang, P., Willems, P. (2015). ‘Combine evolutionary optimization with Model Predictive Control in real-time flood control of
a river system’, Water Resources Management, 29(8), 2527-2542
Wolfs, V., Van Steenbergen, N., Willems, P. (2012), ‘Flood probability mapping by means of conceptual modeling’, River
Flow 2012 (Ed. R.M. Muñoz), International Conference on Fluvial Hydraulics, Costa Rica, 5-7 Sept. 2012; Volume 2, CRC
Press, Taylor & Francis Group, London, UK, 1081-1085; ISBN 979-0-415-62129-8
Breckpot, M., Agudelo, O.M., Meert, P., Willems, P., De Moor, B. (2013), ‘Flood control of the Demer by using Model
Predictive Control’, Control Engineering Practice, 21(12), 1776–1787
Barjas Blanco, T., Willems, P., Chiang, P.-K., Haverbeke, N., Berlamont, J., De Moor, B. (2010), ‘Flood regulation using
nonlinear model predictive control’, Control Engineering Practice, 18(10), 1147-1157
