This
work presents a new decision making criterion for analysis of multiobjective
optimization problems, based on the evaluation of the optimum solutions
in the Pareto set. This new technique allows for discrimination among
the solutions that constitute the Pareto set, so that a single alternative
(the least sensitive one) is selected for real implementation. In
order to do that, the effects of process uncertainties on the variable
responses are analyzed individually, with the help of the process
model, and used to compute a sensitivity index. The proposed technique
was applied to three distinct problems, and the obtained results showed
that the method is able to discriminate the solutions successfully,
allowing for appropriate selection of the best candidates for real
implementation among the many candidates in the Pareto set.