A parallel computational approach for energy-efficient hydraulic analysis of water distribution networks using learning automata

Abstract

The hydraulic analysis of water distribution networks (WDNs) is crucial for ensuring efficient management of water resources, a key aspect of sustainable urban development. Formulation and steady-state hydraulic analysis of these networks have been conducted using both numerical and non-numerical methods. WDN hydraulic equations are complex and non-linear, requiring multiple executions, making their hydraulic analysis computationally demanding and energy intensive. This paper introduces an energy-efficient parallel computing approach using learning automata to significantly enhance the speed and energy efficiency of hydraulic analysis. By employing a cellular automaton framework that reflects the WDN structure, and a solution methodology based on the Taylor series enhanced with learning automata, we propose a system that reduces computational time and energy consumption. We compare the performance of our proposed approach with the EPANET software across networks of varying complexity and topologies. The results suggest our parallel algorithm not only accelerate the hydraulic analysis process up to 60 times compared to existing methods, but also significantly decrease the energy consumption, highlighting its potential for sustainable water management practices.