The goal of this research is to understand how temperature, solvent quality, solvent amount, and the concentrations of organic photovoltaic (OPV) components determine active layer morphology. This understanding will improve techniques for engineering OPV devices, which can be inexpensively processed from abundant materials but presently suffer from low photoconversion efficiencies. We perform molecular dynamics (MD) simulations using HOOMD-Blue accelerated with graphics processing units (GPUs) to quantify how individual molecules self-assemble into structures that influence power conversion efficiency. We simulate blends of poly(3-hexylthiophene-2,5-diyl) (P3HT) with [6,6]-phenyl-C61-butyric acid methyl ester (PC61BM) and [6,6]-phenyl-C71-butyric acid methyl ester (PC71BM), three of the most important molecules in OPVs. By screening hundreds of combinations of concentration, temperature, and solvent properties, we can identify the conditions that optimize their self-organization. We quantify the degree of order in the predicted morphologies with radial distribution functions, structure factors, and simulated diffraction patterns. We find morphologies in agreement with prior experimental and theoretical work, and offer suggestions for future combinatorial studies.