This paper presents a two-level strategy for stochastic synthesis of chemical processes under uncertainty with a fixed degree of flexibility by using the mixed-integer nonlinear programming (MINLP) approach. The objective is to develop an automated and robust strategy which could handle nontrivial optimization problems (about 1000 equations and variables with a considerable number of uncertain parameters - up to 30) that, at the present, cannot be solved in a reasonable period of time by using rigorous stochastic optimization methods. To accomplish the task, the nonlinear subproblems at fixed structures have been decomposed into design and operating optimization levels, the former being facilitated by using a direct search method and the latter by using the reduced dimensional stochastic procedure. Two examples are presented to illustrate the robustness and efficiency of the proposed strategy at solving medium- and large-scale problems.