Ocampo Session 2

HYBRID MPC APPLIED ON SEWER NETWORKS: 

HYBRID MPC APPLIED ON SEWER NETWORKS Carlos Ocampo-Martínez SAC - UPC (Spain) The Barcelona Case Study

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

Barcelona and its rain: 

Barcelona and its rain INTRODUCTION AND SYSTEM DESCRIPTION

New plans: Tanks construction: 

New plans: Tanks construction INTRODUCTION AND SYSTEM DESCRIPTION

Actual Network: 

Actual Network INTRODUCTION AND SYSTEM DESCRIPTION Virtual tank concept. Topology: combined.

Network Objectives: 

Network Objectives Improvement of the Environment Reduction and control of the spills of the sewage system to the receiver (sea and rivers), specially in case of rain. Reduction of the risk of floods Reduction and control of the frequency of the floods and of their effects. Optimization of the operation Optimal advantage of the existing facilities, protection of affected infrastructures and improves of the conservation. INTRODUCTION AND SYSTEM DESCRIPTION

Operation in Real time: 

Operation in Real time INTRODUCTION AND SYSTEM DESCRIPTION Local control  PID Global control  local control references

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

Available Models : 

Available Models Used by CLABSA (Company of residual water management of Barcelona). Network simulators. Highly nonlinear equations. Collectors models (laminar and turbulent flows). Valves and pumps nonlinear models. Alternative linear models and linearization of some dynamics. SYSTEM MODELLING

Linear model: 

Linear model Virtual tanks: water in a catchment (including all set of collectors). Linearization of volume-flow conversion. Linearization of level-flow conversion (Manning). SYSTEM MODELLING Virtual tank 1 Virtual tank 7

RTC of sewer networks: 

RTC of sewer networks M. Gelormino and N. Ricker, “Model-predictive control of a combined sewer system,” International Journal of Control, vol. 59, pp. 793–816, 1994. M. Pleau, F. Methot, A. Lebrun, and A. Colas, “Minimizing combined sewer overflow in real-time control applications,” Water Quality Research Journal of Canada, vol. 31, no. 4, pp. 775 – 786, 1996. M. Marinaki and M. Papageorgiou, Optimal Real-time Control of Sewer Networks. Springer, 2005. M. Marinaki and M. Papageorgiou, “Nonlinear optimal flow control for sewer networks,” Proceedings of American Control Conference, vol. 2, pp. 1289–1293, 1998. M. Marinaki and M. Papageorgiou, “Central flow control in sewer networks,” Journal of Water Resources, Planning and Management, vol. 123, no. 5, pp. 274 – 283, 1997. W. Schilling, B. Anderson, U. Nyberg, H. Aspegren, and P. Rauch, W. Harremöes, “Real-time control of wasterwater systems,” Journal of Hydraulic Resourses, vol. 34, no. 6, pp. 785–797, 1996. M. Schütze, D. Butler, and B. Beck, Modelling, Simulation and Control of Urban Wastewater Systems. Springer, 2002. M. Schütze, T. To, U. Jaumar, and D. Butler, “Multi-objective control of urban wastewater systems,” in Proceedings of 15th IFAC World Congress, 2002. M. Papageorgiou, “Optimal multireservoir network control by the discrete maximum principle,” Water Resour. Res., vol. 21, no. 12, pp. 1824 – 1830, 1985. M. Marinaki and M. Papageorgiou, “Rolling-horizon optimal control of sewer networks,” Proceedings of the IEEE International Conference on Control Applications, vol. 1, pp. 594–599, 2001. S. Duchesne, A. Mailhot, and J. Villeneuve, “Global predictive real-time control of sewers allowing surcharged flows,” Journal of Environmental Engineering, vol. 130, no. 5, pp. 526 – 534, 2004. SYSTEM MODELLING

Case Study: 

Case Study SYSTEM MODELLING

Case Study (II): 

Case Study (II) SYSTEM MODELLING

Case Study (III): 

Case Study (III) SYSTEM MODELLING

Hybrid Elements: 

Hybrid Elements Real tank Virtual tank Collector overflow Redistribution gate SYSTEM MODELLING

Hybrid System Model: 

Hybrid System Model MDL form (Bemporad, Morari, 1999) HYSDEL (Torrissi, 2003) Hybrid Toolbox (Bemporad, 2004) SYSTEM MODELLING

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

System Sub-dynamics: 

System Sub-dynamics Real tanks (physical constraints) Redirection Gates Tank and collector overflows (including the overflows to the sea) Virtual tanks (soft constraints) Maximization of flows (treatment plant) Restricted control signal variation New flow paths (no always present) Hybrid models  MIPC (Mixed Integer Predictive Control)  HMPC CONTROL PROBLEM FORMULATION

HMPC Control Objectives: 

HMPC Control Objectives Minimize CSO in streets (virtual tank overflow). Minimize CSO in links (collectors overflow). Minimize losses to the environment (pollution). Maximize sewage treatment. CONTROL PROBLEM FORMULATION

HMPC Cost Function: 

HMPC Cost Function CONTROL PROBLEM FORMULATION The HMPC problem is then:

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

Simulation of Scenarios: 

Simulation of Scenarios Duration: 6 hours (120 samples) with rain peaks of 55 minutes (11 samples). Prediction horizon: 30 minutes (reaction time of the system to control actions and disturbances). Rain prediction (Gelormino and Ricker, 1994) Known over the horizon. Constant. Rain episodes occurred in Barcelona between 1998 and 2002. RESULTS

Main Results: 

Main Results RESULTS

Main Results (II) : 

Main Results (II) RESULTS

Main Results (III): 

Main Results (III) RESULTS

Main Results (IV): 

Main Results (IV) RESULTS

Outline: 

Outline Introduction and System Description System Modelling Control Problem Formulation Results Conclusions and Further Work

Conclusions: 

Conclusions Hybrid modelling and control of a sewer network has been presented and discussed. Good results using heavy rain scenario. Hybrid modelling not only for nominal behaviours, i.e., faulty mode (Ocampo-Martinez et al, 2005). Main contributions hybrid modelling of a sewer network problems of HMPC implementation on this kind of systems.

Further Work: 

Further Work Prioritization of control objectives (Ocampo-Martinez et al, 2006). Large scale sewage system (Ocampo-Martinez and Bemporad, 2006). Hybrid model of parameters variation. Rain prediction (Radars, stochastic models, statistics studies).