Due to their simple geometry and design, planar heterojunction (PHJ) solar cells have advantages both as potential photovoltaics with more efficient charge extraction than their bulk heterojunction (BHJ) counterparts, and as idealized interfaces to study basic device operation. The main reason for creating BHJs was the limited exciton diffusion length in the active materials of the PHJ: if an exciton is generated at a distance greater than its diffusion length from the hetero-interface of the PHJ, it would be very unlikely to be able to contribute to the photocurrent. Based on this argument one expects a maximum in the photocurrent of PHJs for a thickness of the active layer equal to the exciton diffusion length (~10 nm). However, in two recently developed PHJs that have appeared in the literature, a maximum photocurrent is observed for 60-65 nm of poly(3-hexylthiophene) (P3HT). In this work, we explore this anomaly by combining both an optical T-matrix and a kinetic Monte Carlo simulation that tracks the exciton behavior in the PHJs. The two systems considered are a P3HT/single walled carbon nanotube (SWNT) device, and a P3HT/phenyl-C61-butyric acid methyl ester (PCBM) device. The model demonstrates how a bulk exciton sink can explain the shifted maximum in the P3HT/SWNT case, whereas in the P3HT/PCBM case the maximum is mainly determined by PCBM molecules interdiffusing in the P3HT upon annealing. Based upon the results of this model it will be possible to more intelligently design nanostructured photovoltaics and optimize them toward higher efficiencies.