You could probably make a modification to a weather file for an individual day if all you are comparing is the models sensitivity to a sunny and cloudy day. The modification would be made to the direct and diffuse radiation values and simply running the file and comparing outputs. This would be the simplest way I can think of immediately.