OSSA – GA technology benchmark
Optimization of real-time operations in water distribution systems is challenging due to the large number of parameters in the system and its non-linearity. In addition, for real-time operations the optimization process must be fast enough to reach feasible solutions in short time periods (usually less than 10 minutes).

A common approach is to simplify the network and represent it as a set of mass balance equations and create approximations for pump working points. These approximations can be solved by various linear programming approaches. However, during this process knowledge about the network is lost as the simplified solution does not solve physical hydraulic equations, thus the results do not account for pressure limitations, water quality constrains, etc. 

A more detailed approach is to couple a hydraulic solver with the algorithm that allows for full network simulation, which gives more insight into the operations of the network. With this approach the limiting factor is the run-time of the hydraulic solver, which may limit the applicability of the network-algorithm coupling for real-time operations. A common example of this approach is the incorporation of network solvers with Genetic Algorithms (GA). Peak-Dynamics propose a solution based on an Optimized Stochastic Search Algorithm (OSSA) that enables reaching optimal solutions of complex networks within the time frames required for real-time operation decisions. OSSA was designed and researched by Peak-Dynamics. It is a stochastic search algorithm that starts from a random or given schedule and continues its search in the vast space by adaptively changing and controlling various search parameters. Each candidate solution is tested by a hydraulic simulator and if validated serves as a departing point for the next search. The convergence of this algorithm is exceptionally fast, thus the algorithm can be coupled with a hydraulic solver and still be able to reach optimal solutions within operational time periods. Specifically, the OSSA algorithm was coupled with the EPANET hydraulic solver and has shown promising results in its ability to find optimal operations with large number of optimization parameters (daily and weekly 15 minutes pump schedules for networks with tens of pumps). 

In comparisons performed between the coupled OSSA-EPANET approach and a similar system coupling GA and EPANET, for networks with tens of pumps, the OSSA algorithm converged into optimal (or nearly optimal) solutions by order of magnitude faster.

OSSA – GA benchmark results




For full details download the benchmark presentation

Loading ...