Flexibility analysis is an important task for the optimal design and synthesis of chemical processes with uncertainty. It is a challenging problem because of the discontinuity and nonconvexity of rigorous flexibility programming formulations. In this article, we propose a new parallel hybrid algorithm based on stochastic search in conjunction with a nearest constraint projection technique to numerically solve the flexibility index problem. The proposed method can be applied regardless of the convexity of the design constraints. The stochastic method robustly identifies the global solution without the need for derivative information. The new nearest constraint projection technique is used to handle the constraints of the flexibility index problem in reduced state space. In contrast to existing methods, this technique does not require the addition of artificial variables for active constraints, does not need to have access to explicit analytical forms of the problem formulation or its derivatives, and does not solve for additional artificial variables. Its implementation is well-suited for parallel computing so that computational time can be dramatically reduced. Five applications illustrate the efficacy of the proposed method.