To appropriately select control nodes of a large-scale network system, we propose two controllability centralities called volumetric and average energy controllability scores. The scores are the unique solutions to convex optimization problems formulated using the controllability Gramian. The uniqueness is proved for stable and unstable cases which include multi-agent systems. We show that the scores can be efficiently calculated by using a proposed algorithm based on the projective gradient method onto the standard simplex. Numerical experiments demonstrate that the proposed scores can correctly capture the controllability of each state node compared with existing controllability centralities and the proposed algorithm is better than an existing interior point method from the computational complexity viewpoint.