MINLP model for optimal biocide dosing and maintenance scheduling of seawater cooled plants

Nowadays many countries in arid regions around the world suffer from severe water shortages; this situation is steadily becoming a global concern as water resources are being depleted. In this regard, strategies that a few decades ago were not feasible are becoming interesting alternatives; one of them is the use of seawater as cooling fluid in chemical plants. This alternative, however, presents problems in the plant operation, including a rapid biofilm growth (biofouling) that reduces the heat transfer efficiency. To control the biofouling, the use of biocides (usually chlorine) and periodical mechanical maintenances are implemented. In this work a mathematical programming model is presented for the optimal planning of biocide dosing and mechanical maintenance scheduling. The model considers the biocide kinetics along the network, its dependence with temperature and its interaction with the biofilm thickness. The model uses disjunctive programming to select the optimal maintenance scenario, allowing for different levels of biofilm thickness that depend on the plant schedule. The model also includes environmental constraints to prevent the discharge of streams with potentially dangerous loads of chemicals. These constraints include a limit for the concentration of biocide and the end of pipe chemical treatment. The objective function is to minimize the total annual cost, including the cost for the biocide, chemicals used for end of pipe treatment and maintenance. To show the model applicability, a case study for an integrated power/desalination plant cooled with seawater is solved. Several scenarios were tested including daily, biweekly, monthly and bimonthly biocide and end of pipe treatment chemical dosing, and mechanical maintenances from one to up to six times per year. The results indicate that recurrent dosing policies, either biweekly or monthly for operational simplicity, along with a maximum of three treatments per year provide an optimal scenario.