How to size a solar array for off-grid: step-by-step

Most off-grid sizing mistakes come from stopping one step early: an array that covers the daily load but cannot refill the battery, or one sized in July for a house that is short in December. This guide builds a complete answer in order, using one example throughout. Every load, panel and sun figure in the example is made up to show the method. The EG4 numbers (efficiencies, battery energy, inverter limits) are from the maker’s sheets. For your own numbers, use solar by ZIP and the system finder.

Step 1: daily energy, including the inverter’s own appetite

Energy is power over time: a kilowatt-hour is one kilowatt for one hour. List each load in watts and hours, as in how much battery do you need, and total it. For the example, say the loads add up to 12 kWh per day.

Then add the inverter. The 12000XP lists idle consumption under 70 W, and 70 W for 24 hours is 70 × 24 ÷ 1,000 = 1.68 kWh. Our arithmetic: 12 + 1.68 = 13.68 kWh per day. (See how to read an inverter datasheet for the conditions on that figure.)

Step 2: find the worst month, and its output per kilowatt

Penn State’s stand-alone sizing lesson says such a system should supply loads in any month, and that the month with the highest design ratio, lowest insolation against highest load, is the critical design month. Get the numbers from PVWatts, a tool for estimating the energy production of a grid-connected PV system, which reports monthly and annual totals. Enter your address and a 1 kW system, and read the lowest month.

Peak sun hours are the shorthand. The University of Arizona Cooperative Extension says a location with 3.5 peak sun hours reaches 1,000 W/m² for 3.5 hours per day. That is a measure of energy, not of how long the sun is up. PVWatts already includes a default total system loss of 14% (soiling, shading, mismatch, wiring, connections and more).

For the example, suppose PVWatts shows 2.4 kWh per day from each kW of panels in the worst month. That is invented, though it is what 3.0 peak sun hours would give at about 80% overall efficiency (3.0 × 0.80 = 2.4, our assumption). Your own figure will differ, and for a sunny region may be much higher.

Step 3: add the losses PVWatts does not know about

PVWatts models the array and an inverter. It does not know that your system stores energy and then converts it again. Penn State’s off-grid array formula includes the battery’s efficiency for this reason. For EG4 gear, take the figures from the sheets: the 314Ah battery lists 95% round-trip efficiency, and the 12000XP lists 90.2% maximum efficiency from battery to loads. Both are maxima.

Assume (our assumption) that half the daily energy is used straight from the panels and half goes through the battery. Energy that goes through the battery takes 1 ÷ (0.95 × 0.902) = 1.167 kWh of generation per kWh used (our arithmetic). So the generation needed is:

Step 4: array size and panel count

14.82 ÷ 2.4 = 6.18 kW. With 400 W panels (an assumed size), 6.18 ÷ 0.4 = 15.4, so 16 panels, 6.4 kW.

Now stress it. The PVWatts manual says its errors may be as high as ±10% for annual totals and ±30% for monthly totals against typical weather, and an actual year can deviate by up to ±20% annually and ±40% monthly. If the worst month comes in 30% low, output per kW is 2.4 × 0.7 = 1.68 kWh, and the array needed is 14.82 ÷ 1.68 = 8.8 kW, which is 23 panels. That gap between 16 and 23 panels is the cost of being wrong about one month.

Step 5: battery for the no-sun days, then recharge

Decide how many days the battery alone must carry the house. Penn State calls this autonomy, and ties the days you need to your peak sun hours and the availability you want. For the example, two days.

1. Battery energy out per day: 13.68 kWh ÷ 0.902 = 15.17 kWh (our arithmetic, treating all energy as coming through the battery-to-loads path).
2. Two days: 2 × 15.17 = 30.3 kWh.
3. The 314Ah battery lists 12.86 kWh of usable energy (80% recommended depth of discharge). 30.3 ÷ 12.86 = 2.36, so three batteries, about 38.6 kWh usable. Three batteries are 942 Ah, above the 12000XP’s recommended minimum of more than 400 Ah.

So far the plan looks complete. Now ask the question most plans skip: how long does the battery take to refill? After two sunless days it is down about 30.3 kWh. To put that back at 95% round-trip efficiency takes about 30.3 ÷ 0.95 = 31.9 kWh of generation (our simplification). The refill rate is what the array makes in a day minus what the house uses that day. Here is the example at three array sizes, on an average worst-month day (2.4 kWh per kW):

The array that “covers the load” never refills the battery in winter. Recharge time, not the daily load, is what pushes the array up. The same arrays produce far more than the house needs in summer, when the extra output has nowhere useful to go once the battery is full.

Step 6: the generator as backup

A generator turns the long-recharge problem into hours of run time. The 12000XP sheet lists a recommended generator capacity of 6,000 to 15,000 W. A separate AC charger can also do the job: the Chargeverter GC lists a maximum output of 5,120 W at 240 VAC. Our arithmetic for the example: 31.9 kWh ÷ 5.12 kW is about 6.2 hours to refill the battery at the charger’s maximum, ignoring other losses and any charge-rate limits.

That gives a decision rule. Pick the array for the recharge days you can accept in your worst month, then use the generator for sequences of bad weather longer than your battery days. The 6.4 kW array in the table would need a generator for nearly every refill. The 9.6 kW array refills in about four days, so a generator only covers unusual weather.

Step 7: check the inverter can take the array

Every inverter lists a recommended maximum solar input and a current limit per MPPT. These come from the sheets:

Apply the example. The 12.8 kW array fits the 12000XP, 18kPV and FlexBOSS21 solar limits and exceeds the 6000XP’s 10,000 W. Even the 9.6 kW array fits the 6000XP’s solar limit, but the 6000XP’s 6,000 W output has to carry your biggest simultaneous load, and its recommended battery capacity is more than 200 Ah. Next check strings. The Voc rises in the cold, so run the cold-weather voltage check in the string calculator and read cold weather and solar. Split the panels across MPPTs as the sheet’s current limits allow and compare each string’s short-circuit current with the MPPT’s short-circuit rating on the sheet (44 A per MPPT on the 12000XP). Wiring, protection and the interconnection are electrical work: they need a licensed electrician and local code.

If your largest loads include a well pump or an air conditioner, check surge capability before you pick the inverter: why a well pump trips your inverter. For a model-by-model comparison, see EG4 inverters compared.

How sensitive is the answer?

Three inputs move the result most, and you can test them with the same arithmetic.

• When the energy is used. If all 13.68 kWh went through the battery, generation needed is 13.68 × 1.167 = 15.96 kWh a day, an array of 15.96 ÷ 2.4 = 6.65 kW, or 17 panels. If all of it ran straight from the panels, it is 13.68 ÷ 2.4 = 5.70 kW, or 15 panels. Moving laundry and pumping to the sunny hours saves panels.
• The load itself. At the 50/50 split, each kWh a day you remove saves 1.083 kWh of generation, which is 1.083 ÷ 2.4 = 0.45 kW, about one panel. A cheaper way to shrink the array is often a more efficient appliance, not another panel.
• The worst-month figure. It sits in the denominator, so a 30% error moves the array from 16 to 23 panels, as step 4 shows. This is the number to take from PVWatts for your own address and not from this page.

The example leaves out several smaller things: wiring losses between the array and the inverter beyond the 14% default, the battery’s state-of-charge floor, the temperature derating of the battery, and any change in your loads from season to season. They matter less than the three above, but they all point the same way: toward a bigger array or a longer generator run.

The whole example on one page

Common mistakes

1. Sizing from an annual average. The worst month decides.
2. Forgetting the inverter’s idle draw. It is 1.68 kWh a day at 70 W.
3. Stopping at the battery. If the array cannot refill the battery, the days of autonomy are used once and not replaced.
4. Using the default snow and soiling inputs. PVWatts’ default total loss is 14% with snow at zero. Adjust it for your site, or add margin as in step 4.
5. Counting the maximum efficiencies as typical. They are the best case on each path.
6. Ignoring the inverter’s limits. Check solar input, MPPT current and output on battery before ordering panels.

Next

Put your own numbers into solar by ZIP for the worst month, then let the system finder narrow the hardware. If you are not sure off-grid is the right design, read off-grid, hybrid or grid-tied, and when you are ready to buy, buying solar equipment online.