Our major goal is a "cool roof" using a white PVC membrane to replace our black rubber roof. The secondary goal is a solar panel array that generates energy from direct sunlight as well as reflected light from the cool white roof.

We will need to strengthen the structure of the 77 Huron roof to support an array of 40+ solar panels. But the new PVC roof with the solar panels will be only about half the weight of the old two-layer roof.

The current EPDM rubber roof is past its expected life. We have maintained it by resealing the seams and adding a strip of new rubber on top of the original seams. It was put down on top of the original and very heavy built-up 3-ply (BUR) roof - roofing paper, asphalt (tar/bitumen), and gravel.

Since we will be adding approximately 5000 pounds of weight (new PVC roof - 1500, solar rack mounts - 1000, and solar panels - 2500), we must first tear off the existing roof layers. The tar and gravel roof weighs about 9,000 pounds (600 lbs. per "square"), and the EPDM and Insulation weighs 900 pounds (60 lbs. per square). So we will actually cut the roof weight in half. The distributed weight is only about 3 pounds per square foot compared to the snow load of about 40 psf.

We need to take off the roofs for other reasons. We may need a plumber to inspect the roof drain pipe and perhaps strengthen it and add a new screen to catch leaves.

We also need to properly insulate the "attic" space below the decking, maybe install a thermostatically controlled ventilation fan that exhausts only summer heat. We may have to replace any weakened planks in the decking, possibly even sister bad joists.

And above all, we need to secure the decking planks to rafters/joists with 3.5" #10 deck screws.

When the decking/sheathing is exposed, we can screw the planks into the joists, making them much stronger than nails for holding down solar panel racks.

IronRidge's tech brief shows the relation between panel angle, the south and north tilt legs, and the separation (X) of panels to prevent or at least minimize panel shadowing.

We want to understand how the new roof is fastened to the decking, and how the solar panels will be fastened to the new roof through the PVC to the decking/joists. Do we first need to install anchors to the roof decking/joists/rafters for the solar panel racks, and then install the PVC roofing?

Perhaps not. It may be easier to put on the PVC roof and screw the IronRidge FRAs through the new PVC. There would be no need to cut the PVC, just drive screws through it. We then would hot air weld the IronRidge premolded membrane over the roof attachments to the new white PVC roofing. Installing the tilt leg base would then seal the roof.

The IronRidge diagram above shows an angle of solar incidence of 20°, with no shadowing of the panel behind, even in winter. 20° was recommended by some of our proposed suppliers. We might increase it to 30° or more, which shadows panels some in the winter, but increases the energy collected at all other times, more than compensating for the shadowing loss?

We need to know the values for the panel widths (W = L x cosine tilt angle) and separation (X). Only then can we know how many rows of panels will fit on the roof. We also need to know the reduction of power output when a part of the panel is shadowed.

Solar panels on sloped roofs typically lie in the same plane and never shade one another.

Flat roofs with multiple rows of panels are quite different. In the winter, when the sun is low in the sky, the panel is best when it has a steep angle (65° up from flat). But at such an angle it severely shadows panels behind it. The time when the panel needs to be most upright (winter) is when it most shadows panels behind it.

Panels optimally should be tilted to be perpendicular to the sun's rays. They should normally be fixed at their latitude (for us 42°). Reduced production at the winter and summer solstice is just the cosine of 23°, 92 percent of the maximum at the equinox.

But panels tilted at 42° shade those behind at the winter solstice, when the sun is low in the sky, at elevation 19°.

We would get shadowing from September to March with this angle. At the winter solstice the shadowing is most severe.

Panels at the latitude angle 43° will shadow almost half of the panels behind in winter!

At the winter solstice the angle of solar irradiance for us is 20°. Some installers recommend a 20° tilt angle. The separation that eliminates shadowing (X in the IronRidge diagram above) is 93% of the panel (≅ cos 20°) or about 40" for 40" panels in landscape orientation.

We conclude that (as long as shading does not harm the PV modules) tilting at 43° gives us optimal performance from March to September, when the sun's rays are attenuated least by slanting through the atmosphere.

But production at the winter solstice is reduced greatly for shaded panels, depending on how shaded cells drop the production of the panels.

Losses may be quite severe. Older panels could reduce MPPT (maximum power point tracking) to near zero if even 10% of their cells were shadowed.

Even the best panels have shadowing loss much larger than the shadowing area, depending on how the panel's individual cells are wired.

The table shows best efficiency is achieved with a tilt angle of 20°. The north side of the panel is 13.5" higher than the south. A separation of 36" lowers the shadow at the panel behind enough to eliminate shading completely.

Separating the rows of panels by three feet would eliminate shadowing. IronRidge south tilt legs are available in only one size - 15". North legs are available up to 30". (Which north leg achieves a 20° tilt?)

We could reduce the separation between rows until shading of panels behind is limited to the bottom third of the panel. The bypass diode in the lower string will knock out one-third of the PV panel power, but only for a few weeks around the winter solstice, when irradiance is only one-third of the summer. The power lost in winter is a fraction of the gain in the summer from a 20° tilt, especially when winter power is likely compromised by snow on the panels.

The table value of 70% efficiency in winter only includes the projection effect. It does not include the absorption of energy in the long slant path of solar radiation in the winter. Winter solstice irradiance is only one-third of summer, so the winter power is .7 x .33 = 23%. Losing another one-third of this to shading gives us only 15% efficiency in winter. A few weeks of snow-covered panels would reduce it to single digits!

We can tolerate shading in winter to 33%. But we must guarantee shading affects only one of three strings protected by a bypass diode.

These panels can generate up to 30% more energy using reflected light from the roof surface. Our white PVC roof will be perfect for this.

How exactly do the LG panels deal with shadowing of individual cells? What is the power output when they are shaded 50%? That depends on which part is shadowed. Cells are arranged 6 x 12 wired in series as three strings along the long (12) dimension. In portrait orientation, all strings are affected. In landscape, only the actual strings shaded are affected. We can and must limit shading to the lower one-third in landscape orientation.

These LG panels are more expensive than many others, so tradeoffs must be considered. But in the long run, this is an investment with a very high payback in future years, so maximizing output is probably the best strategy.

We started with Enphase microinverters (which invert panel DC to synchronized AC), but have been advised that they may not handle the power output of the NeON bifacial (390W). This is incorrect. The Enphase IQ 7+ Micro™ (≈$140) is designed for 72-cell panels between 350W and 440W.

SolarEdge power optimizers condition the power with DC-to-DC conversions sent to a single large DC-to-AC inverter.

SolarEdge power optimizer modules, e.g., the SolarEdge P500 (≈$80) for each panel are less expensive than microinverters, but the single inverter (SolarEdge SE100K Inverter, ≈$4000) is expensive, erasing savings on the DC_DC optimizers, and has a shorter lifetime (10-12 years) than panel modules and microinverters at each panel (25 years).

Crane in the garages courtyard will not require a police detail on the street.

Repair bad planks and add deck screws when roof is open.

We need to strengthen the connection between the decking/sheathing and the rafters/joists, which were built using only nails. We need strong deck screws to lock planks into the joists.

We must also renovate the storm drain pipe. Replace it with a sump bowl large compared to the current entrance hole, which has been greatly reduced by old roofing entering the 4" cast iron pipe.

Thickness? 45 mils minimum. 60 mil preferred.

The IronRIdge Flat Roof Attachments (alternatively, (Anchor Products U-Anchors or now OMG PowerPlates Plus) expect to be attached directly to the roof decking.

If attachments were mounted on top of the membrane through 2" of flexible insulation, the screws would bend back and forth and eventually break under the roof membrane?

So it makes more sense to attach FRAs through the PVC membrane, essentially directly to the decking.

Hot air welding of a PVC membrane around roof penetrating objects can be very time-consuming. (Compare labor-intensive FiberTite pipe seals and Johns Manville field-wrap pipe seals.)

By comparison, installing the IronRidge flat roof attachment (FRA) is much quicker and simpler.

1) The IronRIdge FRA is attached with 16 screws (on top of the new PVC roof).
2) The premolded membrane is centered on the FRA by inserting a bolt into the FRA, then hot air welded to the PVC (The 60 mil membrane for Carlisle PVC/KEE is FRA-M60K-CA-W1).
3) The tilt leg is bolted, with weathertight washers, into the (non-penetrating) hole in the FRA.
There is no time during racking installation that the roof has openings for water penetration.

Permit process first.

Specify locations and install the flat roof attachments.

Hot-air weld the PVC/KEE membrane covers for the FRAs. Who does the- roofers or solar installers? Roofers - for warranty reasons.

Install solar rack mounting system on top of new roof.