Author(s): Marcus Nobrega Gomes Junior; Marcio Hofheinz Giacomoni; Eduardo Mario Mendiondo; Ahmad F. Taha
Keywords: Real-time operation; Smart urban water management; Control theory; 2-D Routing; 1-D Routing
Abstract: Flood control in stormwater systems is typically performed by static operations in hydraulic devices such as gates, pumps and tunnels. Along with the development of non-expensive sensors, wireless communication, microprocessors and microcontrollers, newer opportunities to enhance flood management using control theory become evident. Control theory, although has been applied to combined sewer systems, and drinking water systems , , is a relatively new technique in urban drainage infrastructure systems , . Urban drainage facilities as reservoirs, channels, and storage nodes are generally expensive and require useful land. Decision-makers can adapt the existing facilities to meet future climate change and urbanization conditions by properly controlling the flow in the drainage systems through real-time control approaches. The objective of this project is two-fold: (1) develop a linearized extended Kalman-Filter state-space representation of the rainfall-runoff process into (i) cells, (ii) reservoirs and (iii) channels, and (2) compare the efficiency of control strategies obtained by an optimizationbased approach and rule-based approaches in a hypothetical study case. We compared the results of a non-convex nearoptimal model predictive controller solved as a non-linear optimization problem using the pattern search method with rule-based controls of Passive Control (i. e., all valves opened), Detention Control (i. e., valves fully opened after 6-h of the end of the inflow) and On/Off Control (i. e., valves are fully opened if a water level of 3.0 m is reached). A 25-yr, 12hr followed by a 10-yr, 12-hr design storm for San Antonio - Texas was tested in a hypothetical case study.