Dual-purpose desalination plants. Part I. Optimal design

In this paper, a methodology for optimization of a given system configuration for dual-purpose desalination plants will be presented. The whole system is represented as a non-lineal programming (NLP) model solved using general algebraic modeling system (GAMS) [1]. The rigorous model incorporates a high number of non-linear restrictions; so the achievement of the optimal solution is difficult. It is important to point out that the initialization of variables are very important, especially to guarantee the convergence and the determination of the optimal solution. The proposed methodology in this work consists on the resolution of simplified model in order to provide the initial values and critical bounds to solve the rigorous model. In this way, a procedure for the optimization of a given configuration involving the total annual cost (TAC) minimization is presented. The results obtained from one study case applying the methodology are analyzed.

In this paper a methodology and a rigorous nonlineal programming (NLP) model for the optimal design of gas/back-pressure steam turbine cycles coupled to MSF desalination system have been presented. The proposed methodology involves the resolution of a simplified model in order to provide the initial values to solve the rigorous model. Generally, the advantages of using a simplified model are that they provide initial values without computational difficulties, permitting us to solve a more rigorous model efficiently. In various proposed cases computational difficulties arose when different initialization procedures were employed (avoiding the simplified model as a preprocessing stage). For many cases very near optimal solutions were found, for several other cases poor solutions were produced. So, the complete procedure, namely the preprocessing stage and the full problem solution as a final step must be carried out to achieve a good solution.