Do you have another suggestion for considering shading?
Unfortunately HOMER cannot model shading very precisely. Your approach would work quite accurately if HOMER allowed you to specify direct and diffuse radiation in each time step, since shading affects the direct radiation much more strongly than the diffuse radiation. But HOMER does not allow you to specify the direct and diffuse radiation separately. Instead, it asks you to enter the global radiation, which is the sum of direct plus diffuse. When you reduce this number because of shading, HOMER assumes that the reduction is caused by cloudiness, and it adjusts the direct and diffuse radiation differently than it would if it knew about the shading.