Optimization framework based on a sensitivity analysis for the identification of the critical design variables

In this work, a systematic framework was developed and implemented to optimize large scale processes, through a particular case study. In this process, 32 operating and design variables were found, which can be subject to modification using an optimization process. The identification of high impact variables was carried out through a sensitivity analysis using the method of standardized regression coefficients, together with the Latin hypercube method. Once these variables were identified, the optimization of the process was carried out for the maximization of profits together with the minimization of the operating costs using only the most important identified variables. The solution of the process model was done using Aspen Plus and the optimization was performed using the technique of genetic algorithms, which is available in MATLAB. The results illustrated that was possible to increase the profit by 5.02 % and decrease the energy cost 4.6 %. The framework allowed to reduce the computational time 34.2 % compared when all the manipulating variables were used in the optimization task.