Limitations of 3D shade modeling for PV systems

I'm trying to model performance of a flat plate PV system that is shaded by a wall to the southeast. The simulation is showing poorer performance than I expected.

I see a note in the "Add active surface objects" help section that says the active surface object is intended to represent a small rooftop array or subarray and that it is not designed to represent a row in a large roof or ground mounted array that has many rows.

In this case the modeled array is 3 rows of 14 modules and I now have the first row as one active element and is subarray 1. Rows 2 and 3 are both part of subarray 2.

If I'm doing this wrong, any advice on how it should be modeled for accurate results?