Table of Contents

Shade affects PV power output in a way that is not a fixed percentage of the nameplate. Light at a module is the sum of the beam from the sun, the diffuse light from the sky, and the light reflected from the ground. An obstruction blocks the beam on the cells it covers. Diffuse light and reflected light still arrive. A cloudy day, by contrast, is mostly diffuse everywhere. That is a different loss, and it is written up separately in solar panels on a cloudy day. Shade is local. Cloud is the sky.

Because cells in a series string share one current, a hard shadow on part of one module can pull down more of the string than the shaded area suggests. NREL’s shade-database work describes that loss as nonlinear and specific to the configuration. Predictions in that work landed within about 1 to 2 percent of measured data for several systems. The diffuse fraction while shaded was modeled from 10 percent to 100 percent of the plane-of-array irradiance, in 10 percent steps. There is no single “shade cuts output by X percent” that survives that range.

How shade affects PV power output on a string

A tree, a vent, a parapet, or a row in front of another row all cast beam shade. The hour and the month move the shadow. A layout that is clear at noon in June can be covered at 4 p.m. in December. SAM, NREL’s system model, separates beam shade from a single annual sky-diffuse shade loss. Treating “the roof is a bit shaded” as one annual derate hides which hours are gone. Those hours may be the expensive ones on the tariff, or they may be hours the load did not need. The bill, not the module, decides which loss matters.

Module-level electronics change the arithmetic by letting shaded and unshaded modules operate more independently. NREL’s 2016 update of the products it evaluated found annual recovery of shading losses of about 25 to 35 percent. The main difference between the trials was the number of parallel strings, not the equipment brand. Read that sentence carefully. The recovery is a share of the shading loss. It is not 25 to 35 percent more energy than an unshaded array. An unshaded string does not gain that recovery, because there is no loss to recover. On a simple open roof, the extra electronics have less to do.

Bypass diodes inside a module give the current a path around a shaded group of cells. They limit the damage and the heat in the shaded cells. They do not make the shaded group produce. A module with a diode conducting is a module that has given up part of its voltage. That is why a small shadow can still be a large electrical event, and why standing on a module to “check the shade” is a good way to crack cells for no measurement.

How shade affects PV power output differently by mount

On a roof, the shades are usually objects that were there before the array: chimneys, a neighbor’s tree, the building next door. Pacific Northwest National Laboratory’s point about roof access applies to fixing them. You cannot trim what you do not own, and you should not design a string through a shadow you can already see at survey. Ground mounts trade those buildings for row-to-row shade, which is a spacing choice, and for vegetation, which is an operations choice. PNNL lists vegetation and erosion among the ground-mount checks. Grass that shades the bottom of a row is shade you mow, not shade you buy electronics for.

Bifacial modules add a rear path. For a one-sided module, ground-reflected light is typically only 1 to 2 percent of irradiance, so shade on the ground barely matters. For a bifacial module, albedo and the openness behind the row are part of the energy. A near row, a wall, or a dark roof behind the module shades the rear. The front-side shade article and the rear-side article should not be collapsed. Rear gain on one NREL tracker validation was about 6.5 percent annual energy for PERC and about 9 percent for silicon heterojunction, on that configuration. Block the rear and that validation no longer describes your rows. See bifacial solar panels.

Snow is a full-cover shade that the snow model treats as its own loss, not as a tree. Fixed-tilt annual snow losses in a national SAM study ran from about 0 to 16 percent in the continental United States for the designs modeled, with some Alaska sites as high as about 40 percent. Sliding, not melting, was the removal the model relies on. Do not confuse a snow month with a shade defect. The snow page is snow on solar panels.

What to change when shade affects PV power output

Move the module, or move the object, before you buy a more complicated inverter. A vent that shadows the same three modules every afternoon is a layout error. Electronics can recover part of it. They will not recover the part of the day when the beam is simply not on the cell. If the array is already built, log the hours. A monitoring system that shows string current, rather than only site energy, is what makes a shaded string visible. What that system is for is in what a solar monitoring system is.

Do not quote a universal shade loss in a proposal. Quote the hours the survey found, the tools used, and whether any module-level recovery is assumed. If recovery is assumed, say that it is a fraction of the beam-shade loss and that NREL’s range for the products it tested was about 25 to 35 percent of that loss, sensitive to how the strings are paralleled. A proposal that says “shade optimization adds 30 percent” has changed the meaning of the study.

Frequently asked questions

Does a little shade on one solar panel shut off the array?

It can pull down the string that panel belongs to, because series cells share current. It does not, by itself, shut off a different string. How far the loss spreads depends on diodes, module-level electronics, and how the strings are paralleled. The loss is nonlinear. A small shadow is not a small, proportional cut.

Will microinverters or optimizers erase shade?

They can recover part of the shading loss. In the 2016 NREL update that recovery was about 25 to 35 percent of the loss for the equipment evaluated, and string count mattered more than brand. They do not create the beam the obstruction is blocking.

Is shade the same problem in the morning and at noon?

No. The shadow moves. A survey at one hour misses the hour that actually hits the modules. Model the beam shade by hour, and keep diffuse shade as its own term, the way SAM separates them.

Sources and references

Nonlinear, configuration-specific shade loss, agreement within about 1 to 2 percent on several systems, and the diffuse-fraction steps: NREL and University of Colorado shade database. Recovery of about 25 to 35 percent of shading losses for the module-level products evaluated: NREL shading update, 2016. Vegetation and access: Pacific Northwest National Laboratory. Ground-reflected share for one-sided modules: NREL albedo primer.