While single pump design allows operation at high performance levels, multi-pumping systems are controlled to cover actual flow rate requirements, with less emphasis on power consumption. Widely accessible sensors and programmable controllers can improve this approach. This paper discusses the methodology for optimizing analysis of a complex pumping system with set of parallel centrifugal pumps. For this purpose, an experimental model pumping station, with four pumps in a parallel configuration, was designed and constructed. Auxiliary equipment allows control of each pump by three possible methods: (i) discharge valve, (ii) by-pass flow, and (iii) variable speed drive (VSD). For system performance optimization, three estimating strategies, that are functions of control input parameters at each pump were proposed: (i) minimization of power consumption, (ii) flow rate balancing and (iii) maximization of overall efficiency. A Uniquely developed (C++) genetic algorithm (GENOCOP) determined solutions and these were positively verified by measurements at the model experimental station. Results show the existence of multiple local extrema for all of estimating strategies. These solutions show it is possible to control a complex pumping station with various combinations of control parameters, giving similar results. This analysis demonstrates that the strategy of minimizing power consumption is the most energy efficient and proves the described methodology as a powerful tool for optimizing of complex pumping stations. (C) 2016 Elsevier Ltd. All rights reserved.