The WFIRST-AFTA 2.4 m telescope will provide in the next decade the opportunity to host a coronagraph for the imaging and spectroscopy of planets and disks. The telescope, however, is not ideal, given its obscured aperture. Only recently have coronagraph designs been thoroughly investigated that can efficiently work with this configuration. Three coronagraph designs, the hybrid Lyot, the shaped pupil, and the phase-induced amplitude-apodization complex mask coronagraph (PIAA-CMC) have been selected for further development by the AFTA project. Real-world testbed demonstrations of these have just begun, so for now the most reliable means of evaluating their potential performance comes from numerical modeling incorporating diffraction propagation, realistic system models, and simulated wavefront sensing and control. Here we present the methods of performance evaluation and results for the current coronagraph designs.