Minimization model of cost associated with water consumption in soft drink industry

In Brazil, some industries with high consumption of water are developing optimization strategies to prevent or minimize the impact of potential new cost on spring water. The model obtained for the minimization of the water consumption and its operation cost in a soft drink manufacturing facility is NLP and no convex one. This paper shows the results of the application of General Bender's Decomposition algorithm, to a superstructure optimization problem. The use of the GBD algorithm allow nevertheless a global optimal solution which outcome was a water distribution structure with less cost than the one obtained with a traditional NLP solver -local optimum at higher cost. The solver was GAMS (R) / with BDM-LP and Minos (R). A global optimum with the values for a structure, that allows a 17% cost reduction, was obtained in less than one second.