By Paul Grana, co-founder, Folsom Labs
It’s considered common knowledge that you want to point your solar modules south, toward the equator (assuming you are in the northern hemisphere). This maximizes the energy production over the course of the year, through both summer and winter.
Sometimes, however, the homeowner will want to add modules on the north-facing roof. This may be for aesthetic purposes, or sometimes because the south-facing rooftop isn’t fit for solar. The most common rule-of-thumb is that you simply can’t do that. But we wanted to ask, how bad is it to put solar panels on a north-facing roof?
How much worse are north-facing solar modules?
We start with a typical residential system in Charlotte, North Carolina. We designed and modeled the system in HelioScope, our sales and design software platform. With a 2/12 pitched roof (9.5° tilt), the south-facing array will produce 1,361 kWh/kWp . A north-facing array on that same building will produce 1,145 kWh/kWp—a difference of 16% compared to the south-facing array. Not great, to be sure, but probably not as bad as you might expect!
The tilt of the roof matters a great deal. If this same system was on a shallow 1/12-pitched roof (with a tilt of 4.8°), then the south-facing array would produce 1,315 kWh/kWp, while the north-facing array would produce 1,205—a difference of just 8%! If the roof were steeper (say, 4/12), then the north-facing array would be 29% worse.
The orientation of the house also matters. The above examples are for a house facing perfectly north-south. But if the house is facing south-southwest (30° off of perfectly south), then the equator-facing roof is only 14% better. And if the roof is 60° off south, then the equator-facing roof is only 8% better.
We can summarize all of the various roof tilts and orientations for Charlotte in the table below:
Finally, the location also matters, as north-facing modules do better as the array gets closer to the equator. For example, if we were in Florida compared to North Carolina, the north-facing array would be just 12% worse than the south-facing array (versus 16% in North Carolina). On the other hand, that same array in Massachusetts would be 20% worse.
As a rough rule-of-thumb, north-facing modules that are within 10% of the south-facing modules are still extremely likely to be profitable if they can be used to expand the system size (while modules that are within 20% of the south-facing modules are often worth adding). This is because the north-facing modules would incur only the marginal costs (hardware, installation labor) not the fixed costs. After all, you’ve already paid the costs to acquire the customer, to obtain the permits and to send the crew to the site. If the performance gap is smaller than the percentage of fixed costs for the system, then the modules can be profitably added—and in the SAM cost structure, the fixed costs are 32% of residential system costs, implying that the hurdle for profitable inclusion of modules is actually 32%. As a result, we can color-code the various tilt-azimuth combinations for the Charlotte example. The north-facing section of 1/12 roofs are likely to be extremely profitable, while 2/12 rooftops (and select 4/12 rooftops if they are not perfectly facing south) would be worth consideration for the system design. Here are a few examples for Charlotte, Miami and Minneapolis:
Why isn’t energy production as bad as expected?
Most people will be genuinely surprised by these results, with good reason. Pointing modules “away from the sun” is, for many people, something you simply don’t do. But there are a couple reasons for the decent performance of north-facing modules:
Diffuse sunlight will be the same for both the south- and north-facing arrays. There are basically two components of sunlight: the direct beam from the sun (called “direct”), and the glow of the blue sky (called “diffuse”). So while equator-facing modules do better with the direct light from the sun, they both receive similar amounts of diffuse light, which typically accounts for about 30% of the array’s energy.
Direct sunlight is based on a cosine function. The amount of direct sunlight a module receives is based on the cosine of the angle—which, as seen below, is actually somewhat flat, especially near the peak. In other words, the difference between pointing right at the sun, versus being slightly off, is smaller than in other situations.
The sun is overhead in the summer, when the array is most productive—so the arrays are nearly identical during the most crucial times. As can be seen in the chart below, for our original reference project in Charlotte, the north-facing array is nearly identical to the south-facing array in the summer months, when production is greatest. While the differences are much larger in the winter months (over 20%), the energy yield during those times is much smaller.
What about higher-tilted rooftops?
The analysis above shouldn’t imply that roof tilts max out at 4/12, or that north-facing modules always make sense. Moderate roof pitches (from 4/12 up to 9/12) can be common, especially in northern latitudes where the steeper tilt helps the roof to shed snow before the weight builds up. So let’s see how much worse the numbers are for steeper roofs in Minnesota:
In Minneapolis, a 10/12 pitched roof that is perfectly north-south will have a 57% penalty between the south-facing and north-facing modules. In fact, it is pretty unforgiving, even if the house has a southwest angle. And for those in the northeast, Minneapolis is at a latitude of about 45° North, which is in line with Bar Harbor Maine.
On the other hand, steep roofs in Charlotte are less painful, but still not great. A 7/12 pitch could potentially make sense if the roof is oriented to the southwest:
But these modules are clearly borderline (20-30% less productive than the south-facing roof). Clearly, as with any design rule, the right approach will often depend on the location and application.
The rise of north-south rooftops
“Dual-tilt” racking is already popular in commercial flat-roof designs (with products from companies like SunPower, SolarCity/Zep, Everest and Renusol). This new residential design approach would extend those principles to residential systems as well. In an era of cheap modules, ideas that previously seemed crazy can suddenly become completely sound.
 This is a measure of the productivity of an array, providing the number of output-hours the system produces energy. For example, a 5-kW system that produces 1,700 kWh/kWp will generate 8,500 kWh per year (5 x 1,700).
This article was updated on Aug. 3, 2016, to address comments about steep roofs.
Read more Solar Boot-up articles from Folsom Labs here.