To alleviate supply and demand uncertainties, firms have attempted to establish a highly flexible network in their supply chains. This paper proposes to investigate the issues related to process flexibility of production systems facing supply disruption risk and random product demands with differentiated profit margins. To avoid the high cost resulted from supply disruption, we study the associated process flexibility design from a worst-case perspective under a robust optimization framework. In particular, by introducing the marginal profit group index under disruption (MPGID), the worst-case total profit can be represented by the MPGID value of a design and the uncertainty of demands under a partwise independently symmetric perturbation uncertainty set. The achieved representation is beneficial for comparing the worst-case performance of different flexible designs. Based on the proposed model, we further provide a greedy algorithm to quantify the worst-case disruptions. For the case of two products with unequal profits, when supply disruption risk is considered, we prove that the alternate long-chain design is optimal among all types of long-chains. In addition, we also discuss the fragility quantifying the impact of disruptions in the worst case and discover that alternate long chain is more fragile than other long chains if the disruption risk is sufficiently high. Furthermore, an MPGID-based heuristic approach is developed, which can be implemented to generate a flexible design that effectively mitigates the supply and demand uncertainties. Finally, we briefly discuss the disruption of plant node as a special case of arc break.