A sudden failure of a single supporting element within a reinforced concrete (RC) frame can lead to a disproportionate collapse if the design lacks mechanisms to confine initial damage through resisting mechanisms. Given the substantial impact of uncertainties related to material properties and geometric parameters on these mechanisms, coupled with the high stakes associated with such failures, the risk-based optimization offers a practical approach to achieving a balance between cost-efficiency and safety. Besides, optimal risk-based design is strongly dependent on the structural configuration. This study exemplifies this approach by optimizing five RC frames under three scenarios of column removal on the first floor: middle column, penultimate column, and corner column. Design variables encompass cross-sectional depth, steel rebar areas, and concrete strength of beams and columns. Failure consequences are evaluated for both the intact structure and all column removal scenarios. Conducting a physical and geometrical nonlinear static analysis, sample points undergo bay pushdown analysis in OpenSees software. Addressing failure probabilities utilizes the Weighted Average Simulation Method, with risk optimization performed by the Firefly Algorithm. To mitigate computational costs arising from nonlinearities and a high number of required sample points, Kriging is employed to generate an accurate metamodel for limit states and reliability indexes. Results align with the observed trend in Beck et al. [1], justifying progressive-collapse resistant design for larger threats, taller frames, and with optimal distribution of strengthening between beams and columns varying based on the frame's aspect ratio and the column loss scenario.