How much does off-grid solar cost? Battery days, December sun and the DOE 2025 price line

Three formulas set the size, and the size sets the bill

Battery capacity follows from a short line: daily kWh multiplied by days of autonomy gives the energy the battery must deliver, and dividing that by the share of the battery's rating you can actually use gives the nameplate size to buy. A house using 10 kWh a day that wants two dark days covered must deliver 20 kWh from storage. If the battery's spec sheet allows 90 percent of its rating to be used, the nameplate is 20 divided by 0.9, or 22.2 kWh.

Array size follows from a second line: daily kWh, plus an allowance for the energy lost charging and discharging the battery, divided by what one kilowatt of panels produces on an average December day. With a 10 percent allowance and the Denver December figure of 3.10 kWh per kW per day, a 10 kWh household needs 11 divided by 3.10, or 3.55 kW of panels, just to break even on an average December day.

Price is the third line: array watts times a cost per watt, plus nameplate battery kWh times a cost per kWh. This page uses the DOE Q1 2025 residential benchmark for both, because it is the only national, dated, modeled price that covers panels and storage in the same table. A real off-grid bid will differ from it, and the reasons it understates some off-grid costs are set out further down.

None of the three lines needs special software. What makes off-grid design hard is that the December average hides the bad days, and a battery that is exactly one day deep runs out on the first one. The year-long simulation later on shows how often that happens, and what a bigger array or a deeper battery does about it.

Run the storage arithmetic once by hand to see where the $972 comes from. The benchmark's residential system with a 13.5 kWh battery is priced at $4.59 per watt, against $2.95 per watt for the same 8 kW system without one. The difference, $1.64 per watt, applies to 8,000 watts, which is $13,120 for the battery and everything that comes with it in the model. Divided by 13.5 kWh, that is $971.85 per kWh of rated capacity. The benchmark's minimum sustainable price line gives a lower figure: $4.33 less $2.78 is $1.55 per watt, or $12,400, which is $918.52 per kWh.

December output in the two saved PVWatts runs

Two PVWatts runs saved on October 30, 2026 supply the winter numbers. Both model an 8 kW DC array of standard modules on a fixed roof mount, facing due south at a 20 degree tilt, with PVWatts' default 14 percent system losses and a DC-to-AC ratio of 1.2, on the NSRDB PSM V3 typical meteorological year weather files for central Denver and central Los Angeles. The AC output already includes the inverter and wiring losses, which is why this page sizes from it rather than from raw sunlight.

The December figure is what matters off grid. A grid-tied owner banks summer surplus against winter bills, so the annual total is the planning number. An off-grid owner cannot bank anything except what fits in the battery, so every month has to stand on its own, and the weakest month sets the array. In both cities that month is December.

The share in the seventh row is December output divided by one twelfth of the annual total: 768.42 against 1,063.3 kWh in Denver, 801.41 against 1,118.1 kWh in Los Angeles. Los Angeles gets more sun all year, yet its winter penalty is almost the same as Denver's, about 28 percent below an average month. Sizing an off-grid array to the annual average in either city leaves it short every December by roughly that fraction, before any cloudy stretch is counted.

Two cities is a thin sample, and the figures should not be read as a national range. A home farther north, or in a valley with winter fog, will see a lower December number. PVWatts is free to run for any address, and its monthly table gives the December AC output for your own location; divide it by 31 and by the array's kW to get the figure that drops into the array formula. The guide to reading a solar production estimate explains which PVWatts settings matter.

The 20 degree tilt is a typical roof pitch, not an off-grid design. Off-grid arrays are often mounted steeper to face the low winter sun, which raises December output relative to these runs, so the December figures here are conservative for a steeper array. How much steeper helps depends on latitude and snow, and no run at a steeper tilt was saved, so this page does not put a number on it. A ground mount is the usual way to get that tilt, since few roofs offer it.

Battery size and battery price, nine ways

Applying the battery line to three households and three autonomy targets gives the grid of sizes below. The usable share is the 90 percent illustration, and the price is the DOE storage increment of $971.85 per rated kWh. The last two columns count 13.5 kWh units, the size the benchmark models, and check the bank against the residential code's 40 kWh ceiling for batteries inside the house, discussed in the code section.

One row checked by hand: 10 kWh a day for two days is 20 kWh delivered; 20 divided by 0.9 is 22.22 kWh of nameplate; 22.22 times $971.85 is $21,597. The price column is linear in the nameplate, so a battery quoted at your own price per kWh rescales every row by the same ratio.

Batteries come in fixed sizes, so the bank you buy is rounded up to whole units, and the unit column shows how lumpy that gets. A 10 kWh household that wants three days needs 33.33 kWh of nameplate, but three 13.5 kWh units add up to 40.5 kWh, which is both more battery than the arithmetic asks for and just over the indoor limit in the model residential code. Products with other unit sizes round differently; the home battery cost guide breaks a battery quote into its line items so the per-kWh price can be compared.

The usable share deserves its own look. At 90 percent, the 20 kWh delivered needs 22.2 kWh of nameplate. A battery whose maker limits use to half its rating would need 40 kWh of nameplate for the same 20 kWh, and the price line would nearly double unless a lower price per kWh offsets it. Read the usable kWh off the spec sheet, not the headline capacity.

Cycling is the other battery cost an off-grid home pays that a backup battery does not. A battery that carries the house every night cycles every day of the year, and many warranties end at a throughput limit in kWh as well as a number of years, so daily cycling can use up the warranty before the calendar does. The guide to battery warranty throughput shows how to check which limit you will hit first.

Panel count sized to December, priced at the 2025 benchmark's $2.95 per watt

The array line gives the kilowatts needed to cover the December average for each household. Denver uses 3.10 kWh per kW per day and Los Angeles 3.23, both from the saved runs, with the 10 percent charging allowance added to the load. The last column shows what the same households would buy if they sized to the Denver annual average of 4.37 kWh per kW per day instead, which is the mistake examined in the box at the end of this section.

One row checked by hand: 10 kWh plus 10 percent is 11 kWh; 11 divided by 3.0985 (768.42 kWh over 31 days over 8 kW) is 3.55 kW; 3,550 watts at $2.95 is $10,473. The Los Angeles array is only about 4 percent smaller for the same household, because the December gap between the cities is small even though their annual totals differ more.

Two cautions on the price column. The DOE benchmark models an 8 kW rooftop system, and the per-watt price of a 1.8 kW array is unlikely to match it, because fixed costs such as design, permitting and the crew's day spread over fewer watts. Treat the small-system rows as scale, not as a price you should expect. Second, the benchmark is a grid-tied system on a roof; a ground-mounted off-grid array adds racking and trenching the model does not include.

An array sized exactly to the December average still falls short on about half the days of December, because the average includes clear days that produce more than the load and grey days that produce far less. On clear days the surplus beyond what the battery can hold is simply lost. That is why the array and the battery are sized together rather than one after the other.

Sizing panels to the yearly average

An array sized to the annual average looks right on paper: in Denver, 11 kWh divided by 4.37 kWh per kW per day is 2.52 kW, about 29 percent smaller and $3,047 cheaper than the December-sized array. Paired with three days of storage, it still ran dry on 76 days of the typical year and left 199.8 kWh for a generator to supply. Three days of battery cannot make up for an array that falls short on an average day from late autumn to late winter, because the battery seldom gets the chance to refill.

A typical Denver year, hour by hour, for seven designs

The Denver run was saved with hourly output, 8,760 hours of AC production for the typical meteorological year. Scaling that output to each array size and running a battery against a flat household load shows how often each design would run dry. The method is simple enough to repeat: start the year with a full battery; each hour, add any surplus production to the battery at 90 percent of its value, up to the usable capacity; take any shortfall from the battery; when the battery is empty, count the unmet energy and mark the day as a generator day. The table uses a 10 kWh-a-day household; the 5 and 20 kWh households give the same generator days and half or double the unmet kWh, because every quantity scales together.

PVWatts runs on a typical-year weather file, so the year modeled is meant to represent ordinary weather at the site, not its worst; a cloudier year would produce more dry days. A design that shows zero generator days here has no margin built in for a year cloudier than typical, and real homes do not use exactly the same energy every hour.

The cheapest way to reach zero dry days in this year was not the deepest battery. Adding a third day of storage to the December-sized array costs $10,798 and closes the last four dry days. Making the array 50 percent bigger instead costs $5,236 and, with two days of storage, also closes them, finishing $5,562 below the three-day design. Extra panels refill the battery faster on every partly cloudy day, while an extra day of battery only helps during the rare stretch long enough to need it.

The 1.5 times array with two days of storage is not running close to empty either. Its lowest point in the year was 3.58 kWh left of a 20 kWh usable bank, so it came through the worst stretch with a little over a third of a day in hand. The December-sized array with three days bottomed out at 4.25 kWh of a 30 kWh bank.

Doubling the array while keeping one day of storage still left five dry days. One day of storage means the house is empty by the second grey morning no matter how big the array is, because a large array on a heavily overcast day still produces little. Some battery depth is not optional; the question is whether the third day is worth its price.

A 10 kWh cabin priced two ways

Take a 10 kWh-a-day cabin near Denver whose owner wants no routine generator use in a typical year, and price the two designs from the simulation that achieve it.

Design one, deep battery: array 3.55 kW (11 kWh divided by 3.10 kWh per kW per day), costing 3,550 W x $2.95 = $10,473; battery 30 kWh delivered over three days, 33.33 kWh of nameplate at 90 percent usable, costing 33.33 x $971.85 = $32,395. Total $42,868 at the DOE lines.

Design two, bigger array: array 5.33 kW, 1.5 times the December size, costing $15,709; battery 20 kWh delivered over two days, 22.22 kWh of nameplate, costing $21,597. Total $37,306.

Both designs ran the typical Denver year with zero dry days. Design two cost $5,562 less, used a bank of 22.22 kWh rather than 33.33, which fits two 13.5 kWh units inside the 40 kWh indoor limit, and finished its worst stretch with 3.58 kWh in the battery. Divided by 10 kWh of daily use, design two works out to about $3,731 per daily kWh, the figure in the opening of this page, before the off-grid inverter extras, a generator, or any ground-mount work.

The worst stretch in the modeled year came in October

Looking at daily totals for the full 8 kW Denver run shows where the dry days come from. The single worst day was December 15, when the 8 kW array produced 2.9 kWh, about 8 percent of the year's daily average of 34.96 kWh. The worst two consecutive days were October 5 and 6, which produced 19.4 kWh together, 28 percent of two average days. The worst three-day run started on the same day and produced 30.6 kWh, less than one average day spread across three.

December itself swung widely. Its best day produced 35.3 kWh and its worst 2.9, twelve of its 31 days fell below the December average of 24.79 kWh, and three fell below half of it. Across the whole year, eleven days produced under 10 kWh from 8 kW, which is under 1.25 kWh per installed kilowatt.

This is why autonomy days are measured in days and not in months. A battery sized to two days of load matters most in a stretch like the one in early October, when the December average is irrelevant and two days of storage is the difference between a dark house and a working one. Your own location's bad stretches will fall in different weeks; PVWatts hourly output for your address shows them.

Battery days against generator hours

The simulation turns the generator question into a price per avoided kilowatt-hour. Going from the 1.5 times array with one day of storage to the same array with two days costs $10,798 and removes 15 generator days and 26.6 kWh of generator energy in the typical year. That is about $406 of battery for each kWh of yearly generator output avoided. A generator that covers those days instead has its own price: the unit, a transfer switch or inverter input, fuel, and service. EnergyBS prices installed standby generators for the grid-tied case, and the capital side of that comparison applies here too.

The generator also does a job the battery cannot: it covers a year worse than the typical one, a failed inverter, or a week of snow on the panels. For that reason many off-grid designs carry a generator even when the battery math says it will rarely run, and the useful question becomes how small a battery the generator lets you buy. On the table above, accepting four generator days a year by stopping at the December-sized array with two days saves $10,798 against the three-day version.

Maintenance belongs on the same ledger. The DOE benchmark's operations and maintenance figure for the residential system is $34 per kW DC per year without storage and $64 with the 13.5 kWh battery. The $30 difference on 8 kW is $240 a year for the 13.5 kWh unit, or about $17.78 per rated kWh per year. Scaled to the 1.5 times design, that is about $181 a year for 5.33 kW of panels and about $395 for 22.22 kWh of battery, roughly $576 a year in 2024 dollars, before any generator service.

The grid is outside this calculation, and deliberately so. Whether a remote site should pay for a line extension instead, or whether a house with a grid connection should leave it, depends on utility prices and line-extension charges that belong to rate analysis, and EnergyBS's off-grid versus grid-tied comparison takes up that decision. The arithmetic here prices the off-grid system itself.

What the benchmark battery line covers, and what an off-grid build adds

The DOE residential storage case is a grid-tied design: a 13.5 kWh lithium-ion battery cabinet with a 5 kW AC bidirectional inverter, AC-coupled to an array on microinverters. Its $13,120 increment therefore includes a battery inverter, but one built to work alongside a utility connection. The benchmark costs also exclude subsidies payable to the system owner, which makes them a clean comparison line.

An off-grid system has to form its own grid. The inverter must hold voltage and frequency with no utility reference, carry the largest motor start in the house (a well pump is the usual worry), and accept a generator input if there is one. A single 5 kW inverter, the size in the benchmark, may be too small for a house that runs a pump, a range and a dryer at the same time; the inverter's continuous and surge ratings have to be checked against each appliance's nameplate. Some off-grid designs also charge the battery directly from the array through charge controllers rather than through AC coupling. None of these parts has its own line in the benchmark.

Some costs fall away. An off-grid system needs no utility interconnection application, no export meter and no net-metering paperwork, all of which a grid-tied install carries. The benchmark groups office work and other soft costs into broad categories without breaking such items out, so this page does not try to subtract them. The net effect is that the benchmark is a fair scale for the panels and the battery cells, and an incomplete one for the power electronics around them.

The practical use of the line is as a check on a quote. Divide the battery portion of an off-grid quote by its rated kWh and compare it with about $972 per kWh; divide the array portion by its DC watts and compare it with $2.95. A large gap is not proof of a bad quote, since the off-grid extras above are real, but it tells you which line items to ask about.

Numbers to have in hand before an off-grid quote

• Daily kWh for the worst month, from meter data or an appliance list with running watts and hours
• December AC output per kW for your address and tilt, from a PVWatts run
• Days of autonomy you want, and whether a generator will cover anything beyond them
• Usable kWh and round-trip efficiency from the battery's spec sheet, not the headline capacity
• Throughput and year limits in the battery warranty, checked against one full cycle a day
• Inverter continuous and surge ratings against the largest motor load and the combined evening load
• Where the bank will sit, and whether its total kWh fits the 40 kWh indoor or 80 kWh garage and outdoor caps
• Which NEC edition your jurisdiction enforces, and whether it asks for a design review
• The quote's battery price per rated kWh and array price per DC watt, set against $972 and $2.95

Code for a stand-alone system: NEC 690, 706 and 710, and the battery room

Three articles of the National Electrical Code (NFPA 70) carry most of an off-grid installation: Article 690 for the photovoltaic system, Article 706 for energy storage, and Article 710, which Washington's rules title standalone systems. Washington's electrical rules show how a state ties them together. Chapter 296-46B WAC adopts the 2023 edition of the NEC and adopts the 2026 edition effective December 31, 2026, and WAC 296-46B-710 requires a system design review whenever sources regulated under Articles 445 (generators), 480 (stationary standby batteries), 690, 692 (fuel cells), 694 (wind), 700 to 702 (emergency and standby systems) or 706 are interconnected as part of an Article 710 system.

That design review is a set of documents, not a fee line: the manufacturer's installation information, a legible one-line diagram showing the equipment, overcurrent protection, conductor sizing, grounding and any interconnection points, and the calculations used to find the system's voltage and current. Washington requires it to be on site from the first inspection until the inspection process is complete. Other states adopt the NEC on their own schedules and with their own amendments, so the edition your inspector enforces is a question for the local building or electrical department. The guide to DIY solar installation rules covers who may do the electrical work in the first place.

Where the battery sits is a building-code question. The 2021 International Residential Code's Section R328, as reproduced in a Washington State Building Code Council code-change document, requires energy storage systems to be listed to UL 9540, limits individual units to 20 kWh, and caps the total at 40 kWh in utility closets, basements and storage or utility spaces, and at 80 kWh in garages, detached accessory structures, on exterior walls and outdoors on the ground. Banks beyond those ratings fall under Section 1207 of the International Fire Code. Units must be spaced 3 feet apart unless large-scale fire testing documents a smaller distance, and none may go in a sleeping room or a closet opening into one. The 2024 IRC moved the section to R330, and California's 2025 residential code follows that numbering.

Three parts of R328 bear directly on off-grid banks. The 40 kWh indoor cap is the one the battery table runs into: a 20 kWh household with two or three days of storage needs a garage, an outbuilding or an outdoor pad. Section R328.9 requires mechanical ventilation for indoor batteries that produce hydrogen or other flammable gases while charging, so a battery type that gasses while charging cannot go in an unventilated indoor room; the spec sheet says whether yours does. And R328.6 requires inverters listed to UL 1741 or as part of the UL 9540 listing, with utility-interactive listing required only for systems connected to the grid. R328.1 also exempts systems listed to UL 9540 and marked for use in residential dwelling units when installed to the manufacturer's instructions and NFPA 70, so the product's own listing can change which of these rules applies. The battery location rules guide follows each limit through the code editions.

Loads, tilt and snow: adjusting the arithmetic to a real site

Load is the input that moves the total most, and it is the one most often guessed. A month of metered use from an existing house, or a list of appliances with their running watts and hours, gives a better daily kWh than any rule of thumb. Seasonal loads matter more off grid than on: a heat-tape circuit, a block heater or a furnace blower that only runs in winter adds to the December load exactly when production is lowest. If the winter load is higher than the annual average load, size the array to the winter load.

Steeper tilt shifts output toward the winter months, and the summer output it gives up is energy an off-grid home often cannot store anyway once the battery is full. Snow is the opposite case: a snow-covered panel produces almost nothing until it sheds, and a 20 degree roof sheds more slowly than a steep ground mount. The saved PVWatts runs apply one 14 percent loss figure to the whole year rather than a snow history for the site, so a snowy site should treat the December figure as optimistic and lean on autonomy days or a generator.

Shade costs more off grid than on, because a tree that blocks the low December sun removes production in the month that sets the array size. A site that loses its morning or afternoon sun to a ridge or a tree line in winter needs a December figure that reflects that, which PVWatts can only approximate through its loss setting. The guide to solar panels and shade covers how shading reduces output.

Finally, keep the expansion path open. A design that starts with two days of storage and a 1.5 times array can add battery units later if the generator runs more than planned, as long as the inverter, the battery location and the code caps above leave room. Checking those three limits before buying costs nothing and avoids a second permit for a rebuilt battery room.