Solar Array Sizing: Panels or Batteries First, and Why the Answer Isn't Obvious

When an off-grid system runs short, almost everybody adds batteries.

It feels like the obvious fix. The battery is the thing that ran out. Buying another one is a single purchase with no new mounting, no new roof penetrations, and no wrangling with panel angles. The retailers agree with you, enthusiastically, because batteries are the highest-margin component in the system.

In most cases it is the wrong purchase, and the arithmetic is not close.

For anyone whose problem is running out of power on the third or fourth day, array capacity beats battery capacity — usually at lower cost, and by a margin that widens the further you get from midsummer. There are real exceptions, and they are worth understanding properly, but they are exceptions rather than the general case.

Here is the working.

The two things that are actually happening

An off-grid system has a harvest and a load. Batteries sit between them as a buffer.

The battery bank determines how long you can run at a deficit. The array determines whether you are running a deficit at all.

That distinction is the entire argument. Adding battery capacity extends the time you can survive a shortfall. Adding array capacity attacks the shortfall itself. If your daily harvest exceeds your daily load, you can stay out indefinitely on a modest bank. If it does not, no bank is large enough — you have simply bought yourself a longer countdown.

Most people who describe their problem as “I need more battery” are describing a harvest problem.

Why the answer isn’t obvious

The panels-first case is not universal, and three genuine complications keep it from being a slogan.

Roof space is finite and non-negotiable. You can always buy another battery. You cannot always fit another panel. Once the roof is full, array expansion stops being a purchasing decision and becomes a question of ground-deployable panels, which many people will not tolerate the hassle of.

Lead-acid banks cannot absorb unlimited charge. A flooded or AGM bank charges quickly to roughly 80% state of charge and then enters an absorption stage where the battery’s own chemistry limits how fast it will accept current, regardless of how much is on offer. That stage takes hours. Solar systems on lead-acid frequently never reach a true full charge before the sun goes down, which is the root cause of the chronic undercharging that shortens so many lead banks’ lives. Past a certain point, adding panels to a lead-acid system delivers current the battery simply refuses to take.

Batteries solve the night; panels do not. If your consumption is concentrated after dark — heating, lighting, cooking, a satellite dish running 24/7 — you need the buffer regardless of how good your daytime harvest is. A system with a huge array and a small bank browns out at 2am while the panels sit idle in the dark.

None of these overturn the general case. They define its edges.

The arithmetic that settles it

Take a real system. Daily consumption of 2,500 Wh — a fairly typical full-time load once a refrigerator and a permanent internet connection are in the picture, and a figure you can derive for your own setup using the load-calculation method. Existing hardware: 400W of roof-mounted panels and 200Ah of 12V lithium.

It is October. Peak sun hours at this latitude are averaging 3.2. Applying a realistic derate factor of 0.7 for heat, wiring, controller losses and imperfect angle on a flat roof:

Daily harvest: 400W × 3.2 × 0.7 = 896 Wh Daily deficit: 2,500 − 896 = 1,604 Wh

The 200Ah lithium bank holds 2,560 Wh nameplate, of which about 2,050 Wh is usable at 80% depth of discharge. Starting full, that lasts 1.3 days before the system is on its knees.

Now spend money. Two options, roughly comparable outlay.

Option A — add 200Ah of lithium (approximately $600)

Usable capacity goes to 4,100 Wh. The deficit is unchanged at 1,604 Wh/day.

Result: 2.6 days. You bought 1.3 extra days.

Option B — add 400W of panels (approximately $400, plus controller headroom)

Harvest goes to 800W × 3.2 × 0.7 = 1,792 Wh/day. Deficit falls to 708 Wh/day. The original 2,050 Wh of usable storage now drains at less than half the previous rate.

Result: 2.9 days. You bought 1.6 extra days, for two-thirds of the money.

Option C — add 800W of panels (approximately $800)

Harvest goes to 1,200W × 3.2 × 0.7 = 2,688 Wh/day, which exceeds the 2,500 Wh load.

Result: indefinite. The deficit is gone. The bank recharges every day. You can stay out until the water tank runs dry.

There is no battery-only path to Option C. You cannot buy your way to indefinite endurance with storage, because storage does not generate anything. That asymmetry is the whole point, and it is why the reflex to add batteries costs people money.

Where the reversal happens

Run the same exercise in December at 2.2 peak sun hours and the picture changes.

Break-even array for a 2,500 Wh/day load at 2.2 peak sun hours and a 0.7 derate is 1,623W. On a 25ft travel trailer with vents, a fan and an air conditioner on the roof, you will realistically fit 400–800W. On a Class B van, 200–400W. The break-even array does not fit.

At that point the honest answer changes. When the array is physically capped below break-even for the season you actually intend to use, you are no longer sizing for self-sufficiency — you are sizing for endurance between charges, and endurance is what battery capacity buys. Deep-winter and heavily-shaded systems are battery-first systems, supplemented by a generator or by driving, and pretending otherwise leads to a roof full of panels that still cannot close the gap. This is exactly the calculation that sizing a bank for shoulder-season use works through in detail.

The four situations where batteries genuinely come first:

1. The roof is full. Array expansion is physically exhausted and ground deployment is not acceptable.
2. Deep winter or persistent overcast. Peak sun hours below roughly 2.5 with a substantial load, where break-even is unreachable.
3. Heavy shade. Forested sites where harvest is unpredictable rather than merely low. Buffering variance is a storage problem.
4. Generator or alternator is the primary charge source. If the plan is to run an engine for an hour a day, you are optimising charge acceptance, and that is a battery-and-charger question rather than a panel question.

Outside those four, add panels.

Array sizing table

Required array wattage to meet a daily load, at a 0.7 derate factor.

Two things to take from that table.

The first is the spread. The same load requires two and a half times the array in December that it does in June. Any figure quoted without a season attached is meaningless, and most of them are quoted without a season attached — which is why 400W setups routinely disappoint the people who bought them on a summer recommendation.

The second is that you should pick your design season deliberately rather than by accident. Sizing for the annual average produces a system that is oversized for four months, undersized for four months, and correct for none of them. Decide when you actually intend to be out, size for that, and accept generator support outside it.

Realistic roof capacity

Sizing ambitions collide with physical reality faster than most planning admits. Approximate usable roof after vents, air conditioners and antennas:

That last row is why permanent installations should almost never be battery-first. When the array is uncapped, panels are the cheapest kilowatt-hour you will ever buy, and the correct move is to overbuild the array until winter closes on its own.

Tilt, orientation and the capacity left on the roof

The 0.7 derate factor used throughout this guide is not a law of nature. A meaningful part of it is recoverable, and recovering it is cheaper than buying panels.

A flat-mounted array is optimised for a sun directly overhead, which happens for a few weeks a year at midday. The rest of the time the panels are receiving light at an angle, and output falls with the cosine of that angle. In midsummer at mid-latitudes the penalty for flat mounting is modest — perhaps 10%. In October it is 20–30%. In December, when the sun sits low all day and you need every watt, a flat panel can be giving up close to half of what a correctly tilted one would produce.

Tilting a roof array is a nuisance on a vehicle and almost nobody does it. On a permanent installation there is no excuse: a pole or ground mount set at roughly your latitude, or steeper for winter-biased production, is straightforwardly better than a flat roof and usually easier to clean and service.

The related point is orientation. South-facing is optimal in the northern hemisphere, but the curve is flatter than people expect — up to about 30 degrees off south costs only a few percent. What does hurt is parking with the array shaded by the vehicle’s own air conditioner or a roof vent for part of the day, which is a layout problem rather than an angle problem and is worth thinking about before the panels are bolted down.

Shading is not proportional

The single most misunderstood aspect of array behaviour is what happens when part of it is shaded.

A panel is a series string of cells. Shade one cell substantially and it stops conducting, which throttles the entire string behind it — not in proportion to the shaded area, but disproportionately. Bypass diodes limit the damage to a section of the panel rather than all of it, but the principle holds: shading 10% of an array does not cost you 10%.

Where this bites hardest is a series-wired string on a single controller, which is the standard high-voltage configuration recommended for long cable runs. One panel in the shade of a tree branch drags the whole string down. The same panels wired in parallel, or split across separate controller inputs, would have lost only the shaded panel’s contribution.

Two practical consequences. First, if your site or your parking habits involve dappled or partial shade, favour parallel configurations or multiple controllers over one long series string, and accept the heavier cabling that comes with it. Second, when you model harvest for a shaded site, do not model it as a percentage reduction — model it as intermittent, which is a storage problem, and is one of the four cases where batteries genuinely do come first.

The ratio between array and bank

There is a second constraint that catches people who take the panels-first argument too far: the bank has to be able to absorb what the array produces.

For lithium, charge acceptance is rarely the limit. LiFePO4 will typically take 0.5C — a 200Ah bank accepting 100A — which at 12V corresponds to roughly 1,200W of array before the battery becomes the bottleneck. In practice, your charge controller’s amperage rating binds first.

For lead-acid, the constraint is real in both directions. Flooded and AGM banks want a minimum charge current of roughly 10% of capacity to charge properly and avoid sulfation, and they will not usefully accept much above 20–25%. A 200Ah AGM bank wants at least 20A and cannot make good use of much beyond 50A. Above that, extra array capacity is largely wasted — the absorption tail sets the pace, not the panels.

This is one of the quieter reasons lithium has taken over off-grid: it removes the ceiling on how aggressively you can harvest. A lead bank tells you when it is ready to be charged. A lithium bank takes whatever you have.

A workable planning rule: size the charge controller for the array, not the battery, and confirm the bank can accept the resulting current before buying the last panel. The array and battery sizer runs both constraints together and stress-tests the result against a shoulder-season case, which is the scenario that actually breaks systems.

If the installation is permanent

Everything above assumes you are free to mount what you like. On a vehicle, broadly, you are.

On a cabin, outbuilding, barn or any fixed structure, you are frequently not. Ground-mount and roof-mount arrays on permanent structures fall under building and electrical permitting in most jurisdictions, and the thresholds vary enormously — some states exempt small off-grid systems entirely, others require permits, inspection and a licensed electrician regardless of whether the system ever touches the grid. A few counties treat a permitted solar installation as evidence of habitability, which has knock-on consequences for property assessment.

This is worth resolving before ordering hardware rather than after mounting it, because retrofitting compliance is far more expensive than designing for it. The state-by-state comparison of solar permit requirements for outbuildings and cabins sets out where the thresholds sit and which exemptions are real.

The short version

Work out your daily watt-hours honestly. Pick the season you intend to use the system in. Look up the peak sun hours for that season at your latitude, not the annual average. Divide, apply a 0.7 derate, and that is your array target.

Then size the bank for the number of days you want to survive without meaningful sun — one to two days for most people — and check that it can accept the charge current your array will deliver.

If the array target fits, build it, and buy a modest bank. If it does not fit, you are in one of the four exception cases, and the money goes to storage instead.

What you should not do is add batteries because the batteries ran out. That is treating the symptom, and it is the most expensive way to do it.

---