Pump network optimization for a cooling water system

• A pump network structure containing a main and an auxiliary pump networks is proposed.
• A superstructure-based mathematical model is established for the pump network.
• The model is solved by simulated annealing algorithm to minimize the total cost.

Centrifugal pumps are widely used in cooling water systems to transport cooling water to its users. They are installed in the header line of the feed pipe, constituting a main pump network. The pressure head of the main pumps must be large enough to satisfy the pressure heads of all coolers. The pressure drop of parallel branch pipes must be balanced by reducing the opening of valves for some coolers, incurring an energy penalty on some pumps. To attain energy savings, this paper proposes an auxiliary pump network whereby auxiliary pumps are installed in parallel branch pipes. A superstructure-based mathematical model is developed to optimize the total cost of the main and auxiliary pump networks. The optimal number of auxiliary pumps and their installation locations are determined by solving the model with a simulated annealing algorithm. The effectiveness of the model is tested by a case study based on the cooling water network of a refinery.