GII on vertical planes at low latitudes

[JenKH]

I am simulating PV panels in vertical planes (plane tilt 90degrees, e.g. on the sides of buildings). At low latitudes I notice that an azimuth of zero does not give the most GII, instead somewhere to the east or west is better. 

The three screenshots show the effect of azimuth on GII at different latitudes: 
1. Australia, latitude -17deg
2. South Africa, latitude -34deg
3. New Zealand, latitude -47deg

Is someone able to explain to me why the north-facing orientation (in the southern hemisphere) does not have the highest GII for latitudes close to the equator?

Thanks!

[Michele Oliosi]

If you consider the "usual" sun path at equator, basically the sun will rise quite to the east and go up in the sky quickly and then go down to the west. The time spent by the sun around azimuth 0 or -180 is really shortened, with respect to a sun path at higher latitudes, with the sun being lower on the horizon.

Essentially you want to maximize the time your sun is facing the modules, i.e. the azimuth and sun height match the plane orientation. If you are close to the equator, and your POA is north / south or south oriented, the time spent with matching azimuth will be quite low. However with a POA facing east or west the azimuth will match quite well for about half the day.