Sometimes the total (annual) energy harvestable is not the important criterion. On a sunny day in summer you are likely to have excess energy, which is blocked by the controller from reaching the batteries. OTOH in midwinter, or cloudy days, or when there is snow on the panels, you may not fully recharge the batteries. Maximizing the number of days on which the battery gets filled is different from maximum possible energy output.
In my case I started out with 360 watts (two panels) mounted with a fixed orientation at about 45 degrees from vertical (to fit my latitude) and facing about South-SW (to fit the roof). A few years later I added another 410 watts (two more panels), but by then the price of panels dropped a lot, and setting up a full mounting system would have cost about as much as the panels themselves. Moreover, in November and December here it tends to be cloudy most of the time, besides the days being short. On a cloudy day, more so than a sunny day, it is most critical to gather as much energy as possible if the batteries are to be topped off, and a horizontal panel with an open "view" of the cloudy sky can actually collect a bit more energy than one that is facing the (invisible) sun. And in January the panels are likely to be covered with snow. (This is primarily a backup power system, if the grid does go down I can climb the roof and remove the snow.) So I opted to mount the second set of panels flat down on the roof, which has a shallow slope of about 10 degrees, also facing S-SW - far enough from the first set of panels so that they don't shade the second set.
I have the two sets of panels on separate charge controllers. As it happens, the flat-mounted panels seem to generate more electricity than the 45-degree ones early on summer days when the sun first reaches the roof.
