This paper introduces a new formulation of topology optimization for robust design including local failure and load uncertainty. In contrast to most studies, the focus has been on minimizing the total volume with multiple compliance constraints. With the introduction of the reciprocal intermediate variables, the topology optimization problem can be well posed as a sequential quadratic program with exact second-order information. Then, robust topology optimization is implemented within the framework of the suggested formulation, in which not only the randomness of the damage location but also the uncertainty of loading magnitude and direction are taken into account. Finally, several numerical examples are performed to verify the effectiveness and capability of the presented approach for robust design considering local failure and load uncertainty. The effects that varying the input load magnitude and direction, damage location have upon the optimized designs are investigated by comparing those with deterministic design results.