We report simulations of exciton migration and charge separation kinetics in spectrally heterogeneous two-dimensional model antenna/reaction center complexes based on a numerical solution of the Pauli master equation for exciton motion. Experimentally accessible quantities like the charge separation yield, the exciton lifetime(s), the first passage time, the time-resolved and the steady-state fluorescence emission spectra are calculated. The simulated data lend themselves to a direct comparison with experimental data. The diagnostic value of these parameters for studying antenna heterogeneity, antenna structure and trap properties in real photosystems are discussed. Both the 'slow trap' and the 'fast trap' cases are examined in order to simulate the trap-limited and the diffusion-limited kinetics, respectively. For spectrally heterogeneous antenna systems we show that conclusions drawn by other authors regarding the effects of antenna optimisation like, e.g., funnelling, on the charge separation yield and exciton kinetics apply only to the hypothetical fast-trap case. In contrast, in the 'slow-trap case', which is more relevant for natural photosystems, the influence of spectral and structural optimization on the charge separation yield is quite limited. The most relevant parameters that influence the charge separation yield and the exciton kinetics in the heterogeneous antenna case are the ratio of energy transfer rates for trapping vs. detrapping at the reaction center and the spectral heterogeneity of the antenna. We show that the presence of long-wavelength-absorbing pigments (in relation to reaction center absorption) is still compatible with a high yield of charge separation. However, long-wavelength-absorbing pigments will make a pronounced contribution to the steady-state fluorescence emission spectrum, which has considerable diagnostic value. Funneling is shown to be essential for a short first passage time but less important for achieving a high yield of charge separation.