Doc7, you ask a number of good questions
The equation that this method is based on was derived from data that I have from LED lights with rows of LEDs spaced about the same distance as the LED spacing. But, the maximum LED spacing I used was 3.25 inches max., and the number of rows was 3 maximum. Extrapolating from that will work, but probably not for 3X extrapolations. The higher above the substrate the light is, the more you can extrapolate, just because the cones of light get bigger as you go higher. (I think it is about the overlap of the light cones, more than anything else.)
I suspect that with 3 rows of LEDs, 20 inches from the substrate, and 60 degree optics, and 4-5 inch row spacing, you can use n=3. At 20 inches the center half of the cones of light is about 10 inches in diameter, so there is a lot of overlap of the rows of light. I'm not sure that with 40 degree optics you can use n=3, and I suspect you can't. But, for a 20 inch height I wouldn't use 40 degree optics anyway. A 40B tank is about 17 inches high, so, with 3 inches of substrate, the top of the tank is about 14 inches from the substrate, and a light 20 inches away is about 6 inches from the top of the tank. A 40 degree optic would give a circle of light about 4 inches in diameter at the water surface. These circles of light are very visible when they are that small, and wouldn't look good at all.
The PAR numbers you get with those spreadsheet calculation is for right under the middle of the light, so the distance between the outer rows isn't relevant. As with any aquarium light the PAR will drop off near the glass anyway - theoretically. But, light reflected off of the glass adds to the PAR near the glass, greatly reducing the drop off.