Roof, property and home type

Solar panels on a flat roof: 8.3 feet between 10 degree rows at 40 degrees north, and 14.3 kW per 1,000 square feet

At 40 degrees north, 400 watt panels tilted 10 degrees need 8.3 feet from the front of one row to the front of the next to clear the December 21 noon shadow, which fits 14.3 kW per 1,000 square feet of usable flat roof against 18.6 kW for the same modules laid flush on a pitched roof.

The gap between rows is a multiple of the raised edge's height, and the multiple belongs to the latitude. The December 21 noon sun clears a row only if the gap is 1.13 times the height of the raised edge at 25 degrees north, 2.00 times at 40 degrees and 3.16 times at 49 degrees. A rule of thumb that quotes one multiple matches the sun at one latitude only. The sun heights below come from the NOAA solar-position equations, the row geometry is plain trigonometry, and the module is the 400 watt, 1.9 by 1.03 meter composite that the Department of Energy uses in its Q1 2025 residential cost benchmark.

Spacing costs roof area. A house that needs 8 kW and sits under a roof that holds 12 kW at the chosen spacing pays nothing for wide rows, and the best tilt is simply the one that makes the most energy per kilowatt. A house whose roof holds less than its electric bill could use pays for every degree of tilt in lost kilowatts. That second case is the one where tilt, spacing, rounding to whole rows and the shadow a designer is willing to accept all trade against each other, and it is the case the tables here are built for. In that case PVGIS yield runs for four U.S. cities show the flatter row winning every tilt comparison from 5 to 30 degrees, by 5.6 to 23 percent of annual energy, as the break-even sections below work out. No federal tax credit is in any figure.

The December 21 sun sets the spacing

NOAA's General Solar Position Calculations turn a date into the sun's declination in two steps: the day of the year becomes a fractional-year angle in radians, and a short trigonometric series in that angle returns the declination. For day 355, which is December 21 in a 365-day year, the series returns -23.42 degrees. The sun is lowest at solar noon around the winter solstice, so a row that clears its neighbor's shadow at noon on December 21 clears it at noon on every other day of the year.

At solar noon the sun stands due south at a height of 90 degrees minus the latitude plus the declination, which is a negative number in December. That is 41.6 degrees at 25 degrees north, 26.6 at 40 and 17.6 at 49. NOAA's equations place solar noon at 720 minutes minus four times the longitude (in degrees, east positive) minus the equation of time, expressed in UTC, so solar noon is not 12:00 on a wall clock, and the 9:00 and 15:00 used later are solar time as well.

A raised edge of height h casts a shadow that reaches s times h across the roof, where s is the cotangent of the sun's height. The table prints s for each latitude. At 40 degrees north s is 2.00, so a raised edge 1 foot high shades 2 feet of roof behind it at noon. When the sun is off to the side, as it is at 9:00 and 15:00, only the north-south part of the shadow matters for a row that runs east to west, and that part is longer because the sun's height measured in the north-south plane is lower than its true height. At 40 degrees north the sun is 14.0 degrees high at 9:00 but its north-south angle is only 18.5 degrees, which makes s 2.99 and the shadow 49.4 percent longer than at noon.

Sun height on December 21 by latitude, from the NOAA solar-position equations (checked November 4, 2026)
Latitude (degrees north)Sun height at solar noonShadow reach per unit of height at noonSun height at 9:00 and 15:00 solar timeShadow reach per unit of height at 9:00 and 15:00
2541.6 degrees1.1324.8 degrees1.51
3036.6 degrees1.3521.3 degrees1.84
3531.6 degrees1.6317.7 degrees2.30
4026.6 degrees2.0014.0 degrees2.99
4521.6 degrees2.5310.2 degrees4.16
4917.6 degrees3.167.2 degrees5.97

NOAA's own details page for the calculator now carries a notice that the calculator is no longer actively supported or maintained, and the page credits the underlying equations to Jean Meeus's Astronomical Algorithms. NOAA states the sunrise and sunset accuracy as within a minute for latitudes inside 72 degrees. The sun heights here are rounded to a tenth of a degree, which is finer than a row layout can use.

From sun height to row pitch

Take a row tilted at angle beta from the horizontal, with a sloped length L from its low front edge to its high rear edge. Its footprint on the roof is L times cos beta, its raised edge stands L times sin beta above the roof, and the shadow of that edge reaches s times that height. The pitch, measured from the front edge of one row to the front edge of the next, is L times (cos beta plus s times sin beta). The benchmark module standing in portrait has an L of 1.9 meters, 6.23 feet.

At 40 degrees north with a 10 degree tilt, the raised edge is 1.08 feet, the footprint is 6.14 feet and the shadow is 2.16 feet, so the pitch is 8.30 feet. The front edge of the second row sits at the shadow's tip. At noon on December 21 the shadow touches that edge and covers no panel.

Side view of two 10 degree rows at 40 degrees north, set by the December 21 noon sun Two tilted rows on a level roof, south on the left. The dashed noon sun ray (26.6 degrees high) grazes the top edge of row 1 and lands on the front edge of row 2. Footprint 6.14 ft, gap 2.16 ft, row pitch 8.30 ft, raised edge 1.08 ft. Sun, 26.6 degrees highat noon on December 21raised edge 1.08 ftRow 1Row 2footprint 6.14 ftgap 2.16 ftrow pitch 8.30 ftsouthon the left
Side view of two 10 degree rows at 40 degrees north, set by the December 21 noon sun Drawn to scale for a 1.9 meter (6.23 ft) portrait module. Moving the second row closer puts its front edge in the shadow at noon; moving it farther away leaves roof unused. Source: our calculation, NOAA solar-position equations, checked November 4, 2026.
Show the dimensions as a table
QuantityValue
Sun height at noon, December 21, 40 degrees north26.6 degrees
Raised edge of a 10 degree row1.08 ft
Footprint of one row on the roof6.14 ft
Shadow gap2.16 ft
Row pitch8.30 ft

Dividing the sloped length by the pitch gives the share of the roof that modules cover, and that share is 1 divided by (cos beta plus s times sin beta). The length L cancels out. Landscape modules, 1.03 meters on the slope, need a pitch of 4.50 feet instead of 8.30, nearly half, and cover the same 0.751 of the roof. That is why the tables below hold for any module size. Only the whole-row rounding at the roof's edges changes with orientation, and the worked roof further down shows how much that matters on a small roof.

Pitch keeps growing with tilt. At 40 degrees north the noon pitch is 7.3 feet at 5 degrees, 8.3 at 10, 9.2 at 15, 10.1 at 20 and 11.6 at 30. At 30 degrees the pitch is 1.87 times the module's own slope length, so 46 percent of the roof under such an array is gap.

Every distance here assumes a dead-level roof, rows that run east to west and face true south, rows long enough that their ends do not matter, and no parapet or equipment shading the first row. A roof that falls toward the south lifts the sun's angle relative to the roof plane and shortens the shadow along it, while a roof that falls toward the north does the opposite (our geometry, not a quoted rule). A drawing with the real roof angle settles it.

Kilowatts per 1,000 square feet of roof

One thousand square feet is 92.9 square meters. Multiplying that by the module-to-roof ratio and by the module's power density of 204.4 watts per square meter (400 watts over its 1.957 square meters) gives the kilowatts. For comparison, NREL's 2016 rooftop study assumed a ratio of 0.98 for panels on a pitched roof, which allows 1.27 centimeters between modules for clamps, and that ratio puts 18.6 kW on the same 1,000 square feet.

The noon-rule table falls with both tilt and latitude. At 40 degrees north, going from a 5 degree row to a 30 degree row cuts the capacity from 16.2 kW to 10.2 kW, a drop of 37 percent. Latitude compounds the tilt penalty: at 49 degrees north a 30 degree row holds only 7.8 kW per 1,000 square feet, 42 percent of the flush figure, while at 25 degrees north a 5 degree row holds 17.4 kW, 93 percent of it.

Usable roof is the area left after the perimeter margin, the shadow of any parapet, hatches, skylights, mechanical units and the walking paths a roofer or inspector needs. Those deductions come off the roof before any table here applies, and they shape the layout more than the tilt does on a small roof.

Noon rule: row pitch for a 1.9 meter portrait module and kilowatts per 1,000 square feet of usable flat roof, south-facing rows (our calculation, NOAA sun position, 400 watt DOE benchmark module)
Latitude (degrees north)5 degree tilt10 degree tilt15 degree tilt20 degree tilt30 degree tilt
256.8 ft, 17.4 kW7.4 ft, 16.1 kW7.8 ft, 15.1 kW8.3 ft, 14.3 kW8.9 ft, 13.3 kW
306.9 ft, 17.1 kW7.6 ft, 15.6 kW8.2 ft, 14.4 kW8.7 ft, 13.6 kW9.6 ft, 12.3 kW
357.1 ft, 16.7 kW7.9 ft, 15.0 kW8.6 ft, 13.7 kW9.3 ft, 12.7 kW10.5 ft, 11.3 kW
407.3 ft, 16.2 kW8.3 ft, 14.3 kW9.2 ft, 12.8 kW10.1 ft, 11.7 kW11.6 ft, 10.2 kW
457.6 ft, 15.6 kW8.9 ft, 13.3 kW10.1 ft, 11.7 kW11.2 ft, 10.5 kW13.3 ft, 8.9 kW
497.9 ft, 14.9 kW9.6 ft, 12.4 kW11.1 ft, 10.7 kW12.6 ft, 9.4 kW15.2 ft, 7.8 kW
Kilowatts per 1,000 square feet of flat roof at 40 degrees north, by row tilt Horizontal bars. Flush mount 18.6 kW. Noon rule and 9:00 to 15:00 rule by tilt: 5 degrees 16.2 and 15.1; 10 degrees 14.3 and 12.6; 15 degrees 12.8 and 10.9; 20 degrees 11.7 and 9.7; 30 degrees 10.2 and 8.0. 05101520Flush18.65 degrees16.2Noon rule15.19:00 to 15:0010 degrees14.3Noon rule12.69:00 to 15:0015 degrees12.8Noon rule10.99:00 to 15:0020 degrees11.7Noon rule9.79:00 to 15:0030 degrees10.2Noon rule8.09:00 to 15:00kW per 1,000 square feet of usable roof
Kilowatts per 1,000 square feet of flat roof at 40 degrees north, by row tilt Orange bars set the row pitch by the December 21 noon sun, blue bars by the 9:00 to 15:00 sun, each labelled at its end; the grey bar is panels laid flush at the NREL 0.98 ratio. Source: our calculation with NOAA solar-position equations and the DOE Q1 2025 residential benchmark module (400 W, 1.9 by 1.03 m), checked November 4, 2026.
Show the bar values as a table
Row tiltNoon rule (kW per 1,000 sq ft)9:00 to 15:00 rule (kW per 1,000 sq ft)
Flush mount18.618.6
5 degrees16.215.1
10 degrees14.312.6
15 degrees12.810.9
20 degrees11.79.7
30 degrees10.28.0

The tables use true south. Orientation errors cost less than tilt errors, according to the Sandia PV Performance Modeling Collaborative, but the penalty grows as the array is pointed away from south, so a building rotated well off the compass needs its own geometry; the row-pitch formula changes once the rows are no longer perpendicular to the noon sun.

Choosing which winter shadow to accept

The noon rule is the loosest rule that leaves every panel unshaded at solar noon on the shortest day. A designer who wants more winter production sets a window: no shadow on any panel from 9:00 to 15:00 solar time on December 21, for example. Neither window is taken from a code or a standard; they are two labelled choices that show the range a layout sheet might have assumed. The next table recomputes the first with the 9:00 to 15:00 window.

9:00 to 15:00 rule: row pitch and kilowatts per 1,000 square feet of usable flat roof, same module and assumptions
Latitude (degrees north)5 degree tilt10 degree tilt15 degree tilt20 degree tilt30 degree tilt
257.0 ft, 16.8 kW7.8 ft, 15.2 kW8.5 ft, 14.0 kW9.1 ft, 13.0 kW10.1 ft, 11.7 kW
307.2 ft, 16.4 kW8.1 ft, 14.6 kW9.0 ft, 13.2 kW9.8 ft, 12.1 kW11.1 ft, 10.6 kW
357.5 ft, 15.9 kW8.6 ft, 13.7 kW9.7 ft, 12.2 kW10.8 ft, 11.0 kW12.6 ft, 9.4 kW
407.8 ft, 15.1 kW9.4 ft, 12.6 kW10.8 ft, 10.9 kW12.2 ft, 9.7 kW14.7 ft, 8.0 kW
458.5 ft, 14.0 kW10.6 ft, 11.1 kW12.7 ft, 9.3 kW14.7 ft, 8.0 kW18.4 ft, 6.4 kW
499.5 ft, 12.5 kW12.6 ft, 9.4 kW15.7 ft, 7.6 kW18.6 ft, 6.4 kW24.0 ft, 4.9 kW

The window costs more as the latitude rises. Moving from the 9:00 to 15:00 rule to the noon rule adds 6 percent to the capacity of a 10 degree array at 25 degrees north, 13 percent at 40 and 32 percent at 49. At 15 degrees the gains are 8, 17 and 41 percent.

What each hour of clear window costs

At 40 degrees north the choice can be laid out one hour at a time. Each hour of clear window demanded either side of noon pushes the pitch out a little at first and then quickly: the third hour, which takes the sun from 20.7 to 14.0 degrees high, costs 1.1 kW per 1,000 square feet at 10 degrees, while the first hour costs about 0.1 kW (the table rounds 14.26 and 14.13 to 14.3 and 14.1).

What a shadow accepted at the edge of the window looks like on the next row can be computed too. With rows at 10 degrees pitched to the noon rule at 40 degrees north, the shadow of row 1 covers 0.8 percent of the sloped length of row 2 one hour from noon, 3.9 percent at two hours, 11.4 percent at three hours (the 9:00 and 15:00 sun) and 35.6 percent at four hours, when the sun is 5.5 degrees high. Those are shares of the panel's length in shadow at that instant on December 21. They are not energy losses. How much power a strip of shadow along a module's lower edge removes depends on how the module's cells are wired into bypass-diode sections, whether portrait or landscape orientation puts that strip across one section or all of them, and whether panel-level electronics are fitted; the shaded roof guide and the inverter comparison cover those effects, and the annual figure for self-shading comes from the PVWatts graph discussed below.

Latitude 40 degrees north: what each design window costs in roof area, 400 watt module (our calculation)
Rows stay unshaded for this many hours either side of solar noon on December 21Sun height at the edge of the windowPitch and kW per 1,000 sq ft at 10 degreesPitch and kW per 1,000 sq ft at 15 degrees
None (noon rule)26.6 degrees8.3 ft, 14.3 kW9.2 ft, 12.8 kW
1 hour25.1 degrees8.4 ft, 14.1 kW9.4 ft, 12.7 kW
2 hours20.7 degrees8.6 ft, 13.7 kW9.7 ft, 12.1 kW
3 hours (9:00 to 15:00 rule)14.0 degrees9.4 ft, 12.6 kW10.8 ft, 10.9 kW
4 hours5.5 degrees12.9 ft, 9.2 kW16.1 ft, 7.4 kW

What NREL assumed for flat roofs

NREL's 2016 technical report Rooftop Solar Photovoltaic Technical Potential in the United States: A Detailed Assessment (NREL/TP-6A20-65298, January 2016, by Gagnon and colleagues) had to decide what a flat roof holds without drawing any one roof. It defined a flat roof as one with a tilt under 9.5 degrees, a choice "consistent with many roofers' and PV installers' definition of flat roofs". A 2:12 roof slope is 9.46 degrees, so the report's cutoff and the 2:12 threshold in the building code excerpt discussed below nearly coincide; we treat the match as a coincidence.

Its simulation assumptions for flat roofs were a 15 degree tilt, a south-facing azimuth of 180 degrees, a module power density of 160 watts per square meter and a ratio of module area to roof area of 0.70. In the report's words: "For flat roofs, the ratio of module area to roof area was assumed to be 0.7 to reflect the row spacing necessary to incur only approximately 2.5% losses from self-shading for south-facing modules at 15-degree tilt. For tilted roofs, the value was assumed to be 0.98 to reflect 1.27 cm of spacing between each module for racking clamps." The same report found that flat planes are "very common on large buildings (93% of planes on large building are flat) and medium buildings (74%) but less common on small buildings (26%)". Flat planes are the norm on large and medium buildings and the minority on small ones, so a small house with a flat roof is the less common case.

The report does not say for which latitude 0.70 holds, so the following is our arithmetic and not NREL's statement. A ratio of 0.70 at 15 degrees equals the noon-rule ratio at 37.4 degrees north. South of that latitude, 0.70 is wider spacing than the noon rule needs; north of it, the ratio packs rows tighter than the noon rule allows and accepts some shadow at noon. At 40 degrees north the noon rule at 15 degrees gives 0.674, at 45 it gives 0.617, and the 9:00 to 15:00 rule gives 0.575 and 0.489. NREL's single national ratio therefore reads as a mid-latitude design with a small accepted shadow.

Set side by side, the published densities are easy to compare. NREL's 0.70 flat-roof ratio with its 160 watt per square meter module is 10.4 kW per 1,000 square feet. The Department of Energy's 2009 consumer guide Own Your Power says a system "needs about 100 square feet of unshaded south-facing roof or yard space for every kilowatt", which is 10 kW per 1,000 square feet, with the guide adding that thin-film systems may require 175 square feet per kilowatt. The benchmark's 400 watt module is denser than the 160 watt per square meter module NREL modeled, at 204 watts per square meter, so 10 degree rows under the noon rule at 40 degrees north hold 14.3 kW.

Roof density assumptions side by side, kilowatts per 1,000 square feet of usable roof
BasisModule area to roof areaModule power densitykW per 1,000 sq ft
NREL 2016 flat roofs (15 degree tilt, south)0.70160 W per square meter10.4
NREL 2016 tilted roofs0.98160 W per square meter14.6
DOE 2009 consumer guide, 100 square feet per kilowattnot statednot stated10.0
DOE 2025Q1 benchmark module laid flush (NREL 0.98 ratio)0.98204.4 W per square meter18.6
Same module, 10 degree rows, noon rule, 40 degrees north0.75204.4 W per square meter14.3
Same module, 15 degree rows, noon rule, 40 degrees north0.67204.4 W per square meter12.8

How PVWatts handles row shading

The PVWatts website (version 8.7.4 with API version 8.5, checked November 4, 2026) treats row-to-row shading differently by array type, and the difference matters for a flat-roof layout. Its help text for the Shading loss says: "For one-axis trackers, PVWatts calculates self-shading losses using the Ground Coverage Ratio you specify under Advanced Parameters, so you should not use the shading loss to account for self-shading with the one-axis tracking option." For fixed and two-axis arrays it says: "For fixed arrays or arrays with two-axis tracking, you can use the graph below to estimate a self-shading loss percentage when you know the ground coverage ratio (GCR). The GCR is defined as the ratio of the array area to the ground or roof area occupied by the array."

Our reading is that a fixed flat-roof array in PVWatts takes its row shading from the number the user types into the Shading box, taken from that graph, and not from a spacing calculation. The help text does not say what the engine does internally for a fixed array, and the developer page for the API lists a gcr input, default 0.4, described only as "Ground coverage ratio". The default shading entry is 1 percent, which the help describes as "an array with little or no shading". On that reading, a flat-roof run left at that default carries only a 1 percent allowance whatever the packing, so a roof packed at 0.75 needs the graph's number typed in.

That ratio is the same quantity as the module-to-roof ratio computed above, which is how the tables here connect to the graph. NREL's chart, drawn for a latitude of approximately 35 degrees north, plots shading loss against GCR for fixed arrays at 10, 20, 30 and 40 degrees. Read by eye from a small image, the 10 degree curve is at or under about 1 percent until the GCR passes 0.6, near 3 percent around 0.72 and between 6 and 7 percent at 0.79, while the steeper curves leave the floor earlier: the 40 degree curve is near 10 percent at a GCR of about 0.6. These readings carry perhaps one to two points of uncertainty.

At 35 degrees north and 10 degrees, the 9:00 to 15:00 layout has a GCR of 0.723 and the noon layout 0.789. If the chart readings hold, the tighter layout fits 9 percent more kilowatts and gives up roughly 3 to 4 points more energy to self-shading, a net gain of about 5 percent on a roof that cannot hold more. That is an order-of-magnitude check, not a yield estimate; the installer's shading model for the real roof outranks it.

Tilt against yield: the break-even rule

The Sandia PV Performance Modeling Collaborative describes a fixed array by a tilt and an azimuth, and gives a south-facing latitude-tilt array in Albuquerque, at 35.05 degrees north, as its example. Its page on tilt errors shows that a one degree error at latitude tilt changes annual plane-of-array irradiance by only 0.04 percent, and adds that "errors in the tilt angle have a greater effect for horizontal systems than for latitude tilt systems". A flat-roof rack at 5 to 15 degrees is much closer to horizontal than the latitude-tilt case the page describes. The text gives no percentage for a miss of 20 to 30 degrees, because the relation is plotted in a figure and not stated in numbers, so none is taken from it, and the PVGIS runs in the next section supply the yield numbers.

A flatter row has a cost in output per kilowatt and a gain in kilowatts per roof. The two can be compared without knowing either yield. If a steeper row fits only 89.8 percent as many kilowatts as the flatter row on the same roof, it must produce more than 11.4 percent extra per kilowatt to win on annual energy. The table lists that threshold for the usual flat-roof tilt pairs. At 40 degrees north, a 15 degree row beats a 10 degree row on a roof-limited site only if its annual kilowatt-hours per kilowatt come out at least 11.4 percent above the 10 degree row's. The PVGIS runs reported after this section put that gap at 3.3 percent at Denver.

Extra annual output per kilowatt a steeper row must deliver to beat a flatter row on the same roof (our calculation from the density tables)
Steeper tilt against flatter tilt35 degrees north, noon rule40 degrees north, noon rule45 degrees north, noon rule40 degrees north, 9:00 to 15:00 rule
10 against 5 degrees11.4 percent13.8 percent17.0 percent19.7 percent
15 against 10 degrees9.4 percent11.4 percent13.8 percent15.7 percent
20 against 15 degrees7.9 percent9.4 percent11.4 percent12.8 percent
20 against 10 degrees18.1 percent21.9 percent26.7 percent30.4 percent
30 against 15 degrees21.1 percent25.8 percent31.5 percent35.7 percent

The way to use the table is to divide the annual output per kilowatt at the steeper tilt by the output at the flatter tilt and compare the result with the table entry. If the ratio of yields is less than the threshold, the flatter row makes more energy on that roof. If the roof is not the limit, the table is irrelevant and the higher-yielding tilt wins. The PVGIS runs in the next two sections supply the ratios for four cities, and they settle the question for these sites: for every pair in the table the yield gain is 1.8 to 6.7 percent against thresholds of 7.9 to 31.5 percent, so on a roof that limits the system the flatter row wins in all 20 comparisons. At Denver, 15 against 10 degrees needs 11.4 percent and PVGIS gives 3.3 percent (1,537 against 1,488 kWh per kilowatt); the 10 degree roof therefore makes 1.114 / 1.033 = 1.078 times the energy, about 8 percent more. The 5 degree row, the flattest tested, beats the 10 degree row at all four cities. A reader can repeat the test for another address with any production model run at both tilts, entering each layout's self-shading loss from the PVWatts graph if PVWatts is the model; PVGIS and PVWatts are different models and should not be mixed within one comparison.

Snow changes the comparison in cold places. The NREL snow-loss table reproduced in the snow-loss guide prints losses at a 20 degree tilt and at latitude tilt, so a 5 or 10 degree rack has no column of its own. In that guide's examples the loss at 20 degrees is higher than at latitude tilt, for instance 10.7 against 9.5 percent at the Minneapolis station and 5.5 against 3.9 at Boulder, so a snowy site needs its own estimate for a low rack. The PVGIS runs below carry no snow loss, so the margin they leave is the room for it: at Minneapolis, 10 degrees against 5 needs 17.0 percent and PVGIS gives 4.4, so a low rack would have to lose about 11 percent more of its annual energy to snow than a 10 degree rack before the order flipped (1 - 1.044 / 1.170 = 10.8 percent), against the 1.2 points that table shows between 20 degrees and latitude tilt. That table has no 5 degree column, so this is a bound and not a measurement. The same guide covers how to turn a loss percentage into years of payback.

PVGIS yield by tilt in four cities

The break-even test needs the extra output per kilowatt that a steeper row actually delivers. To put numbers on it, annual output per kilowatt of panel was taken at six tilts from the Photovoltaic Geographical Information System (PVGIS), the free tool of the European Commission's Joint Research Centre. The runs use PVGIS 5.2 with the PVGIS-NSRDB radiation database, which the PVGIS documentation lists for the Americas between 60 degrees north and 20 degrees south. That covers every U.S. site used here, and each request returned results for the database's 2005 to 2015 years, so no site was dropped. PVGIS-NSRDB is offered only through the version 5.2 address. On November 4, 2026 the 5.3 address rejected it with an error and accepted only the PVGIS-ERA5 and PVGIS-SARAH3 databases.

Each run is one kilowatt-peak of crystalline silicon, fixed and free-standing, facing true south (azimuth 0 in PVGIS), with the total system loss set to 14 percent. The 14 percent is a figure entered by hand, since the interface has no default. A flat loss scales every tilt by the same factor, so the ratios between tilts do not depend on it. The sites are Albuquerque (35.08 N, 106.65 W), Denver (39.74 N, 104.99 W), Chicago (41.88 N, 87.63 W) and Minneapolis (44.98 N, 93.27 W), picked to sit on or near the 35, 40 and 45 degree columns of the break-even table; Albuquerque is also the city Sandia uses in its example. Each request took this form, with the site in lat and lon and the tilt in angle: https://re.jrc.ec.europa.eu/api/v5_2/PVcalc?lat=39.74&lon=-104.99&raddatabase=PVGIS-NSRDB&peakpower=1&loss=14&angle=10&aspect=0&outputformat=json. The calls were made on November 4, 2026.

PVGIS 5.2 annual output in kWh per kWp for fixed south-facing rows, with the ratio to the 30 degree tilt in parentheses (30 degrees is the best of the six tilts run, not the PVGIS optimum; 14 percent system loss; the last row averages rows facing east and west)
TiltAlbuquerque, NMDenver, COChicago, ILMinneapolis, MN
5 degrees1,659 (0.895)1,428 (0.878)1,111 (0.912)1,156 (0.866)
10 degrees1,721 (0.928)1,488 (0.914)1,146 (0.941)1,207 (0.905)
15 degrees1,771 (0.955)1,537 (0.944)1,174 (0.964)1,250 (0.937)
20 degrees1,810 (0.976)1,577 (0.969)1,196 (0.982)1,286 (0.964)
25 degrees1,837 (0.991)1,607 (0.987)1,210 (0.994)1,314 (0.985)
30 degrees1,854 (1.000)1,628 (1.000)1,218 (1.000)1,334 (1.000)
10 degrees, east and west rows averaged1,572 (0.848)1,349 (0.829)1,058 (0.869)1,091 (0.818)

Output climbs with tilt at every city across the whole range, and the steps shrink as the tilt grows. The step from 5 to 10 degrees adds 3.1 to 4.4 percent; the step from 25 to 30 degrees adds 0.6 to 1.5 percent. A 5 degree row gives up 8.8 to 13.4 percent against the 30 degree row, and a 10 degree row gives up 5.9 to 9.5 percent. Those are the sums a roof with room to spare cares about, because there the tilt that yields most is the right one. Averaging an east-facing and a west-facing row at 10 degrees gives 0.904 to 0.923 of a south-facing row at the same tilt, a figure used in the east-west section below.

PVGIS is a different model from PVWatts. It derives the sunlight on the panel from its own satellite-based database and applies its own temperature and reflection losses, so the kilowatt-hours per kilowatt in the table will not match a PVWatts run at the same address, and the two should not be mixed in one comparison. The ratio between two tilts is the only quantity the break-even test uses, and that is the quantity reported here. Four limits apply. The runs are for south-facing rows only. They contain no snow loss, and the snow-loss table cited above shows more snow loss at 20 degrees than at latitude tilt, so Denver, Chicago and Minneapolis are flattered at low tilts in winter. They contain no shading between rows, so every tilt is treated as an unobstructed array, and they contain no parapet or equipment shadow, only the terrain horizon PVGIS computes for the site. And four cities are not every climate, so a coastal or persistently cloudy site could behave differently. The figures are long-run averages for a typical year, not a forecast for a particular roof.

Flatter rows against the yield loss

Dividing each steeper row's yield by the flatter row's gives the percentage gain that the break-even table is compared with. The table below sets each gain beside the threshold from the column for the site's latitude: Albuquerque against the 35 degree column, Denver against 40, Minneapolis against 45, and Chicago, at 41.9 degrees north, against 40. Thresholds rise with latitude, so using the 40 degree column for Chicago gives the steeper row the lower bar and understates how far ahead the flatter row is. All thresholds are for the noon rule.

PVGIS yield gain of the steeper row over the flatter row, against the gain it needs to beat the flatter row on the same roof (noon rule; our arithmetic on the PVGIS 5.2 runs above)
Steeper against flatter tiltAlbuquerque, NM: gain against neededDenver, CO: gain against neededChicago, IL: gain against neededMinneapolis, MN: gain against needed
10 against 5 degrees3.7 against 11.4 percent4.2 against 13.8 percent3.1 against 13.8 percent4.4 against 17.0 percent
15 against 10 degrees2.9 against 9.4 percent3.3 against 11.4 percent2.5 against 11.4 percent3.6 against 13.8 percent
20 against 15 degrees2.2 against 7.9 percent2.6 against 9.4 percent1.8 against 9.4 percent2.9 against 11.4 percent
20 against 10 degrees5.2 against 18.1 percent6.0 against 21.9 percent4.3 against 21.9 percent6.6 against 26.7 percent
30 against 15 degrees4.7 against 21.1 percent5.9 against 25.8 percent3.7 against 25.8 percent6.7 against 31.5 percent

In all 20 pairs the gain is below the threshold, so the flatter row makes more energy on the same roof. The tightest case is Albuquerque at 20 against 15 degrees. The 20 degree row yields 2.2 percent more per kilowatt (1,810 against 1,771 kWh) but needs 7.9 percent, a shortfall of 5.7 points, and the roof of 15 degree rows makes 1.079 / 1.022 = 1.056 times as much energy, 5.6 percent more. The widest is Minneapolis at 30 against 15 degrees: a 6.7 percent gain against 31.5 needed, and the 15 degree roof makes 1.315 / 1.067 = 1.232 times as much, about 23 percent more.

A Denver roof shows the same thing in kilowatt-hours. Under the noon rule at 40 degrees north, 10 degree rows hold 14.3 kW per 1,000 square feet and 15 degree rows 12.8 kW. At 1,488 and 1,537 kWh per kilowatt, that is about 21,300 kWh a year against 19,700, so the flatter roof makes 8 percent more. Under the 9:00 to 15:00 rule the thresholds in the 40 degree column are higher still, 12.8 to 35.7 percent, against Denver and Chicago gains of 1.8 to 6.0 percent, so the flatter row wins by more.

The conclusion has edges. It covers south-facing rows from 5 to 30 degrees at four cities, from one model. Nothing here speaks for a flush or dead-flat array, where drainage and dirt also matter, because no tilt below 5 degrees was run. It applies only when the roof is the limit on system size; where the roof has room to spare, the steeper row wins, by 10 to 15 percent per kilowatt at 30 degrees against 5. And shading between rows, which PVGIS leaves out, is the main thing that could change the order. The smallest margin is 5.7 points, so the flatter row would need to lose about 5.3 percent more of its annual energy to shading than its steeper neighbour (1 - 1.022 / 1.079 = 5.3 percent) to flip even the tightest pair. The PVWatts graph reading above put the extra self-shading of the tighter spacing at 3 to 4 points at 10 degrees, which is below that margin, but it compares two spacings at one tilt and not two tilts, so it is a check on scale and not a proof. An installer's shading model for the real roof outranks this table.

East-west pairs pack tighter

Instead of facing every row south, a layout can face rows east and west in back-to-back pairs with a small gap between pairs. With a 0.3 meter gap between pairs, which is our assumed value, a 10 degree pair covers 0.940 of the roof, 17.9 kW per 1,000 square feet. A real layout sets that gap from the rack maker's design and the maintenance access the roofer wants, and shading between pairs is not computed here. The 17.9 kW is therefore a ceiling for the pairs and the shares in the table are minimums: a real layout that accounts for shading between pairs holds fewer kilowatts and needs a higher share.

The east-west layout holds more kilowatts, but each kilowatt makes less annual energy because no row faces south. The break-even is again a ratio. At 40 degrees north, south-facing rows at 10 degrees hold 14.3 kW under the noon rule and 12.6 kW under the 9:00 to 15:00 rule, against the pairs' 17.9 kW, so the pairs win on annual energy for a roof-limited house if they yield at least 80 percent or 71 percent per kilowatt of what the south rows yield. PVGIS runs at 10 degrees with the rows facing east (azimuth -90 in PVGIS) and west (azimuth 90) average 0.914 of a south-facing 10 degree row at Albuquerque, 0.907 at Denver, 0.923 at Chicago and 0.904 at Minneapolis. Against the shares in the table (84 and 77 percent at 35 degrees north, 80 and 71 at 40, 75 and 62 at 45) the pairs clear every threshold, by 7.4 points at the tightest, Albuquerque under the noon rule, and by 10.7 to 15.4 points at the other three cities under the noon rule (Chicago read against the 40 degree column), and by 14.4 to 28.4 points at all four under the 9:00 to 15:00 rule. At Albuquerque that is 17.9 kW at 0.914 against 15.0 kW, about 9 percent more energy for the pairs, which is the thinnest edge and the one most easily lost: the 0.3 meter gap is assumed, and shading between pairs, which neither the layout arithmetic nor PVGIS includes, would take some of it. Chicago's east and west rows differ by 7 points (0.957 and 0.888), so one orientation can matter more than the average. The same caveat applies as for tilt: the sums are for these four cities from one model, and the installer's layout decides the real answer.

East-west back-to-back pairs at 10 degrees against south-facing rows at 10 degrees (our arithmetic; 0.3 meter gap between pairs assumed, shading between pairs not computed)
Latitude (degrees north)South rows, noon rule, kW per 1,000 sq ftSouth rows, 9:00 to 15:00 ruleEast-west pairsEast-west must yield at least this share of south per kW (noon rule)Same share (9:00 to 15:00 rule)
3515.013.717.984 percent77 percent
4014.312.617.980 percent71 percent
4513.311.117.975 percent62 percent

Two cautions come from the primary sources. The 2009 DOE consumer guide says performance is about 5 percent less with a southeast- or southwest-facing system and that "eastern, western, and northern exposures are not recommended", but that guide describes roof planes of ordinary pitch and does not measure a ballasted 10 degree pair, so we cite it only as a direction. Sandia's azimuth page says the annual effect of an azimuth error changes with the azimuth angle but "not as much as for the tilt angle", and that "the effect of the alignment error is greatest as the system is pointed away from due south". A low tilt makes the azimuth penalty smaller, because a nearly horizontal surface has little preference for a direction; that is a geometric fact and not a figure from either source. The ground mount versus roof mount guide runs the orientation comparison on a pitched roof for one city.

Fire access rules at 2:12 and below

A pitched roof loses a strip along the ridge and a pathway to fire-code setbacks, and the fire setbacks guide works those out. A low-slope roof can be outside them. Larimer County, Colorado's residential roof layout sheet, updated January 2026, excerpts the 2024 International Residential Code. Section R329.6 there begins: "Roof access, pathways and setback requirements shall be provided in accordance with Sections R329.6.1 through R329.6.2.1." It then lists four exceptions. Exception 1: "Detached, non-habitable structures, including but not limited to detached garages, parking shade structures, carports, solar trellises, and similar structures, shall not be required to provide roof access." Exception 2: "Roof access, pathways and setbacks need not be provided where the code official has determined that rooftop operations will not be employed." Exception 3: "These requirements shall not apply to roofs with slopes of two units vertical in 12 units horizontal (17-percent slope) or less." Exception 4: "BIPV systems listed in accordance with UL 3741, where the removal or cutting away of portions of the BIPV system during fire-fighting operations has been determined to not expose a fire fighter to electrical shock hazards."

Exception 3 measures the slope of the roof and not the tilt of the rack, so a dead-level roof with 15 degree rows is below the 2:12 threshold, which is about 9.5 degrees. This page takes "these requirements" as the roof access, pathway and ridge-setback requirements of R329.6.1 through R329.6.2.1 named in the opening sentence; R329.6.3, the rule about the roof below an emergency escape and rescue opening, is a separate section that the exception's wording does not obviously reach. The county sheet also reproduces a local amendment: within Poudre Fire Authority's jurisdiction, where the authority "has determined that roof top operations for fighting a fire will not be deployed", the setback from the ridge can be reduced to 12 inches for panels on one side of the ridge. That shows how a jurisdiction can change the model text.

So the exception relieves a dwelling's flat roof of the setback layout the code prints for pitched roofs, and it is a statement about one model code and one county's excerpt. A building outside the residential code, a state or city amendment, and an inspector's own view of walkways and access can each bring conditions back. The building department's answer for the address, in writing, decides how much of the roof is available before the pitch table is applied.

Height limits, ballast and roof load

Washington adopts the 2021 International Residential Code (WAC 51-51-003), which numbers its solar section R324 where the 2024 edition excerpted above, as the Larimer County sheet numbers it, uses R329. Its amendment, WAC 51-51-03240 (the section carries filing history through a rule effective March 16, 2024), has a prescriptive path for rooftop systems. Its R324.4 exception says the roof structure "shall be deemed adequate to support the load of the rooftop solar photovoltaic system if all of the following requirements are met", and the list includes: "3. The total dead load of modules, supports, mountings, raceways, and all other appurtenances weigh no more than 4 pounds per square foot (19.5 kg/m2)." and "4. Photovoltaic modules are not mounted higher than 18 inches (457 mm) above the surface of the roofing to which they are affixed." Item 1 requires design for the local wind speed and installation per the manufacturer's specifications, item 2 a ground snow load of at most 70 pounds per square foot, and item 5 supports that spread dead load so that "no point load exceeds 50 pounds (22.7 kg)".

Item 4 does not say where the 18 inches is measured, so our reading is that it caps the height of the module's raised edge above the roof. On that reading a portrait benchmark module reaches 18 inches at about 13.9 degrees of tilt: its raised edge is 6.5 inches at 5 degrees, 13.0 at 10, 19.4 at 15 and 25.6 at 20, with no allowance for a rack standoff. A landscape module of 1.03 meters reaches 18 inches at about 26.4 degrees. Under an 18 inch cap read this way, landscape is the only orientation of this module that reaches a tilt above about 14 degrees.

Item 3 has no stated area basis, so two readings are printed. The Department of Energy's cost model puts module mass at 10.99 kilograms per square meter of module, so one benchmark module weighs 21.5 kilograms, 47.4 pounds, which is 2.25 pounds per square foot of module area. Spread across the roof the spaced array occupies at 40 degrees north and 10 degrees (the noon rule, 28.1 square feet of roof per module), the modules alone load 1.69 pounds per square foot. If 4 pounds per square foot applies to the roof area occupied, each module's share is 112 pounds and 65 pounds remain for rails, supports and ballast. If it applies to module area, each module's share is 84 pounds and 37 pounds remain. The Department of Energy's commercial benchmark describes its ballast blocks as "40-pound blocks typical", so the remaining allowance is on the order of one to two such blocks per module under the first reading and less than one under the second. Whether a ballasted rack's blocks count as "supports, mountings, raceways, and all other appurtenances" is a question for the building department, and how many blocks a particular array needs is set by the rack maker's wind engineering for the site, not by this arithmetic.

The model file behind the Department of Energy's 2025Q1 benchmark carries two mounting cases. The residential 8 kW case is parallel to a pitched roof, with "flashing feet and accessories to mount rails to a pitched roof". Its only ballasted case is the 250 kW commercial school roof, "ballasted tilted roof mounts", with ballast trays (16 by 18 inches typical), concrete blocks, rails and clamps. Summing the component lines on the minimum sustainable price basis (its market-price column adds the Section 301 tariff to rails and clamps), the residential mounting lines come to $35.89 per square meter of module ($0.176 per watt at the benchmark module's density) and the commercial ballasted lines to $38.67 ($0.183 per watt at that case's 211 watts per square meter), with ballast trays and blocks making up 57 percent of the commercial total. The two lines differ by less than a cent per watt. The commercial case also carries a roof-preparation line of $11.19 per square meter, "Cost of roof modifications to accommodate system installation", and the residential case has no such line and excludes roof repair. The two cases describe different building types, so nothing here is a price for a ballasted flat-roof house. Our reading is only that the model shows no large gap in rack hardware per watt between the two mounts, and that roof work and structural review sit in other lines; the roof weight guide walks through the structural side.

One 40 by 30 foot roof, laid out

Ratios hide the effect of whole rows. A 40 by 30 foot flat roof with a 3 foot margin on every side (the margin is our assumption, not a code value) leaves 34 feet east to west and 24 feet north to south, 816 square feet of array area. The flush ratio of 0.98 would put 15.2 kW on that area. Whole panels in whole rows fill it very differently: a row fits as many 400 watt panels as 34 feet divides into panel widths, and the number of rows is the first footprint plus however many full pitches fit in the remaining depth.

The table prints panels and kilowatts for the portrait and landscape layouts, at four tilts and the two rules. At 10 degrees under the noon rule the portrait layout holds 30 panels and 12.0 kW, and the landscape layout 25 and 10.0. At 15 degrees the order flips: portrait 20 panels, landscape 25. Under the noon rule neither orientation wins at every tilt: portrait holds more at 10 degrees, landscape more at 15 degrees and when laid flat, and they tie at 5 and 20. The roof's depth decides which one rounds favorably, and a layout sheet that does not try both may have left kilowatts on the table. Flat-laid panels with no tilt, at 30 panels in portrait and 35 in landscape, do not reach the flush figure because a 24 foot depth is not a whole number of panel lengths.

A 40 by 30 foot flat roof at 40 degrees north with a 3 foot margin all round (34 by 24 feet of array area, 816 square feet; margin is our assumption), 400 watt modules whole panels only
Row tiltPortrait, noon rulePortrait, 9:00 to 15:00 ruleLandscape, noon ruleLandscape, 9:00 to 15:00 rule
Flat-laid (no tilt, no gaps)30 panels, 12.0 kW30 panels, 12.0 kW35 panels, 14.0 kW35 panels, 14.0 kW
5 degrees30 panels, 12.0 kW30 panels, 12.0 kW30 panels, 12.0 kW25 panels, 10.0 kW
10 degrees30 panels, 12.0 kW20 panels, 8.0 kW25 panels, 10.0 kW25 panels, 10.0 kW
15 degrees20 panels, 8.0 kW20 panels, 8.0 kW25 panels, 10.0 kW20 panels, 8.0 kW
20 degrees20 panels, 8.0 kW20 panels, 8.0 kW20 panels, 8.0 kW20 panels, 8.0 kW

The 9:00 to 15:00 rule drops the 10 degree portrait layout from 30 to 20 panels, because a third row no longer fits in the 24 foot depth, a loss of 4.0 kW from one decision about a winter shadow. Three rows at the 9:00 to 15:00 pitch need 24.9 feet of depth, under a foot more than the 24 available, so on a roof this small the whole-row step is the real design constraint.

Roof area an 8 kW system needs

The Department of Energy's Q1 2025 residential benchmark is an 8 kW system of twenty 400 watt modules at $2.95 per watt DC in 2024 dollars, $23,600, with no tax credit in the price and a pitched-roof mount. Turning the density tables upside down gives the usable roof area those twenty modules need on a flat roof. Flush at NREL's 0.98 ratio they need 430 square feet; the 2009 rule of 100 square feet per kilowatt would set aside 800.

At 40 degrees north, 10 degree rows under the noon rule need 561 square feet, 31 percent more than flush, and 633 under the 9:00 to 15:00 rule. At 15 degrees the noon rule needs 625 and the 9:00 to 15:00 rule 733. The numbers are for usable roof, so a roof whose usable share is half its footprint needs twice them in total. The panel sizing guide sizes the system from the electric bill, and the cost per square foot guide shows how the benchmark price converts between panel, roof and floor-plan areas; both feed the flat-roof question once the usable area has been drawn.

The price line itself does not describe a flat-roof house. The Department's benchmark model has no residential ballasted case, and the $2.95 per watt belongs to the pitched-roof system. For a flat roof the right use of the line is to compare a quote's price per watt against it, as for any roof, and to ask for the production model and the structural scope that explain any gap.

Usable roof area (square feet) needed for 8 kW of 400 watt modules (20 modules), south-facing rows (our calculation); flush mount at the NREL 0.98 ratio needs 430 square feet
Latitude and rule5 degree tilt10 degree tilt15 degree tilt20 degree tilt30 degree tilt
30 degrees north, noon rule469513554590649
30 degrees north, 9:00 to 15:00 rule487550608661753
40 degrees north, noon rule493561625684786
40 degrees north, 9:00 to 15:00 rule529633733826994
45 degrees north, noon rule513600683760897
45 degrees north, 9:00 to 15:00 rule5737198619961242

Reading the layout sheet of a quote

A flat-roof proposal that prints a panel count without a row pitch has hidden the one number that explains it. The sheet should show the tilt of the rows, the pitch or the gap between them, the day and hours the spacing was designed to, the panels per row and the number of rows, the margins kept at the roof's edge and around equipment, and the roof area the array occupies.

Holding those against the tables above is quick. Take the latitude, the tilt and the design rule, find the row of the table, and compare its pitch with the sheet's. A sheet pitch well under the noon-rule pitch has accepted shadow at noon on December 21, and the production estimate should say how much it assumed. A sheet pitch well above the 9:00 to 15:00 pitch has spent roof the house may have needed.

The production estimate should state the self-shading loss it used for the layout, and the structural scope should say whether the rack is ballasted, attached or both, what roof load the rack maker calculated for the site's wind and snow, and who stamped it. A quote that answers those has shown its work; one that does not leaves the pitch table as the only check available.

Four items to find on a flat-roof layout sheet

  • The row tilt and the row pitch (or the gap between rows), so the roof coverage ratio can be recomputed from the formula above.
  • The design day and hours of the spacing, whether December 21 noon, 9:00 to 15:00 solar time or something else, and what self-shading loss the production estimate used.
  • The array area and the margins kept at the edge, the parapet and the equipment, and whether portrait and landscape were both tried.
  • The mounting method (ballast, attachment or both), the roof load the rack maker calculated for the site, and the building department's position on any code limit such as a height cap or a dead-load threshold.

Flat roof solar panel questions

How far apart should solar panel rows be on a flat roof?

Far enough that the front row's shadow does not reach the next row at solar noon on December 21, at minimum. For a 1.9 meter portrait module tilted 10 degrees at 40 degrees north that is a pitch of 8.3 feet, a gap of 2.16 feet between the footprint of one row and the front of the next. The gap is 1.13 times the raised edge's height at 25 degrees north and 3.16 times at 49 degrees, so the answer changes with latitude. A designer who wants clear sun from 9:00 to 15:00 needs 9.4 feet at the same latitude and tilt.

What tilt is best for solar panels on a flat roof?

It depends on whether the roof or the electric bill limits the system size. If the roof is the limit, a flat one wins: in PVGIS runs for Albuquerque, Denver, Chicago and Minneapolis, the flatter row made more annual energy per roof in all 20 pairs of tilts from 5 to 30 degrees. At 40 degrees north a 10 degree row holds 14.3 kW per 1,000 square feet and a 15 degree row 12.8 under the noon rule, so the 15 degree row must make at least 11.4 percent more energy per kilowatt to come out ahead. PVGIS gives it 3.3 percent at Denver (1,537 against 1,488 kWh per kilowatt), so the 10 degree roof makes about 8 percent more. If the roof has room to spare, the steeper row wins: 30 degrees yields 10 to 15 percent more per kilowatt than 5 degrees. PVGIS is a different model from PVWatts, covers south-facing rows only, and has no snow loss or shading between rows. NREL's national rooftop study used 15 degrees for flat roofs.

How many solar panels fit on a flat roof?

It depends on the usable area, the tilt, the latitude and the whole-row rounding. At 40 degrees north, 10 degree south-facing rows fit about 14.3 kW per 1,000 square feet of usable roof under the noon rule, which is about 36 panels of 400 watts. On a 40 by 30 foot roof with a 3 foot margin the layout table fits 30 panels in portrait at 10 degrees and 25 in landscape, because rows round down to whole numbers.

Do solar panels on a flat roof need fire setbacks?

Under the 2024 International Residential Code text excerpted by Larimer County, Colorado, an exception says that these requirements do not apply to roofs with slopes of 2 units vertical in 12 units horizontal (17-percent slope) or less, and this page takes those to be the roof access, pathway and ridge-setback requirements. That is an exception in one model code as excerpted by one county. Buildings outside that code, local amendments and the code official's views can change the layout, so the building department's written answer for the address is what counts. The fire setbacks guide covers pitched roofs.

Is an east-west layout better on a flat roof?

It fits more kilowatts but each makes less annual energy. At 40 degrees north our arithmetic puts back-to-back 10 degree pairs at 17.9 kW per 1,000 square feet against 14.3 for south-facing rows under the noon rule, so on a roof-limited house the pairs win if they yield at least 80 percent per kilowatt of what the south rows yield. The pair gap of 0.3 meters is an assumption. PVGIS puts east and west rows at 10 degrees at 0.904 to 0.923 of a south-facing row at four cities, which clears the thresholds of 62 to 84 percent, with the thinnest edge about 9 percent at Albuquerque before any shading between pairs.

Does PVWatts account for shading between rows?

The PVWatts website help says one-axis trackers get self-shading from the ground coverage ratio entered under Advanced Parameters, and that for fixed arrays the user can estimate a self-shading loss from a graph of loss against GCR and enter it as the Shading loss, whose default of 1 percent represents little or no shading. On our reading of that help text, a fixed flat-roof run left at the default carries only that 1 percent for shading; the help does not say what the engine does with the GCR input for a fixed array.

Does any figure here include the federal solar tax credit?

No. The federal residential solar credit is not available for property placed in service after December 31, 2025 (the IRS's wording; the statute says expenditures made after that date), and the DOE benchmark price of $2.95 per watt DC (2024 dollars) cited here excludes any tax credit or subsidy payable to the system owner. The figures here are roof areas, kilowatts and sun angles.

A flat roof gives a designer the freedom to choose a tilt, and the freedom has a price in roof area that the sun's height sets before any other decision is made. The row pitch follows from the latitude, the tilt and the winter shadow a designer will accept; the kilowatts that fit follow from the pitch. Once those are known, the rest is a short list of questions for the quote: what was assumed about winter, what the production model did with the shading, and what holds the array down.

Sources and verification

Row pitches and kilowatts are geometry for south-facing rows on a dead-level roof, with the NOAA sun position for December 21, 2026 and the DOE 400 watt, 1.9 by 1.03 meter benchmark module, not a roof-specific design or a quote. The noon rule and the 9:00 to 15:00 rule are labelled design choices and not code requirements; the 3 foot margin, the 0.3 meter east-west gap and the one to two blocks of ballast are our assumptions and arithmetic. Latitudes of 25 to 49 degrees north are tabulated; other latitudes are not in the tables. The PVWatts shading graph was read by eye from an image. Annual yield by tilt comes from PVGIS 5.2 runs (PVGIS-NSRDB, south-facing, 14 percent system loss, four cities), a different model from PVWatts that leaves out snow and shading between rows and has no soiling term beyond the flat 14 percent system loss; no PVWatts production run is printed. The 2024 International Residential Code is cited from a county excerpt, the Washington rule is a state amendment, and neither is legal or engineering advice; other states and codes are not in the comparison. The NREL rooftop study is the 2016 edition and a newer one may exist. No federal residential solar tax credit is in any figure. Nothing here is engineering or financial advice.