Next-generation systems must be able to fly a variety of vehicle types in multiple mission scenarios, and to manage off-nominal situations that could lead to flight performance degradation, and abort triggering. To reach a satisfactory level of autonomy, Fault Detection, Isolation and Reconfiguration functionalities are integrated to act on all or a part of the G&C chain. However, in the case of strong actuator failures and reduced control authority, the flight controller may not be reconfigured properly and timely so as to track the reference guidance profiles. In this case, the guidance unit may be reconfigured, whether by relaxing the guidance tracking objectives or by replanning and/or retargeting a new feasible or sub-optimal TAEM trajectory. This second option requires the use of a fault-tolerant on-board path planner (FTOPP), which has been retained in the frame of SICVER, a collaborative research project funded by ESA, ASTRIUM Space Transportation and the Regional Council of Aquitaine (France). This project is driven by IMS-LAPS, Université Bordeaux 1 and is part of the VER/VEP project, supported by the Aerospace Valley cluster. While several methods exist in the open literature, the on-board computation of TAEM trajectories still remains a challenging issue. In fact, many of the proposed methods use large approximations of the governing dynamics and/or require a great amount of computational resources to numerically propagate several TAEM trajectories. In this paper, sub-optimal reference TAEM trajectories are computed on-board using an innovative optimization method that combines flatness approach, convexification and nonlinear programming techniques. The main advantage over the existing techniques is that the original optimal control problem can be transformed into a geometric set-up, leading to an integration-free programming problem.