Every undersized battery bank was sized by wishful thinking; every oversized one by fear. Proper sizing is neither — it is a short chain of honest arithmetic, and it starts not with the battery but with the duty: what must keep running, at what power, for how long, how often, and how cold or hot the room gets. Get those five answers right and the ampere-hours almost calculate themselves. This is the chain we walk for every enquiry, whether it ends in a 2 V stationary cell or a traction bank.
Step 1 — The load, in watts, without optimism
List every load the battery must carry and add the watts. Nameplate ratings lie in both directions — motors draw more at start, electronics less at idle — so the honest figure is measured where possible, worst-case where not. Two refinements matter in industrial rooms: loads quoted in VA must be multiplied by their power factor to get true watts, and any load that only runs part of the backup period is counted for its own hours, not the whole window. The output of Step 1 is a single number: watt-hours the battery must deliver — watts × backup hours, summed across loads.
Step 2 — The losses between battery and load
The battery does not feed the load directly; an inverter or converter sits between, and it takes its cut — conversion efficiencies are often taken around 80–90% depending on the equipment and its loading point. The battery itself has an energy efficiency below 100% as well. Divide the watt-hours from Step 1 by these efficiencies (as decimal fractions) and the number grows — that growth is reality arriving. Cable runs to a distant load add their own drop; on low-voltage DC systems even a modest cable resistance steals meaningful volts, which is a sizing input, not an afterthought.
Step 3 — The usable window, not the label capacity
A battery’s label capacity is not yours to spend in full. Two windows shrink it. First, depth of discharge: cycle life falls as discharges deepen, so the designer chooses a working DoD from the datasheet’s cycles-versus-depth curve — the economics of replacement decide, not a slogan. Deep-cycle tubular designs are built to range far deeper than automotive plates, which is exactly why the tubular plate exists. Second, rate: capacity depends on how fast you pull it. The same cell delivers fewer ampere-hours at a 1-hour rate than at its 10-hour or 20-hour rating — the rate effect Peukert quantified back in 1897, explained in the C-rate guide — so the capacity you look up must be the one at YOUR discharge rate: the datasheet’s rate table, not its headline. And the bank must still hold the load’s minimum voltage at the end of the discharge, at the highest current it will see; a bank that meets the ampere-hours but sags below the cutoff five minutes early has failed the sizing.
Step 4 — Temperature, the quiet resizer
Rated capacity is stated at a reference temperature, typically 25–27 °C. Cold rooms deliver less — the chemistry slows — so a bank in an unheated hill-station shelter or a cold store must be sized up using the manufacturer’s temperature-correction factors. Heat gives a little capacity back but taxes it with shorter life, which is a different bill arriving later. Either way: size for the coldest credible operating day, not the average one.
Step 5 — Margin, then the arithmetic
A small engineering contingency — a few percent for ageing, measurement error and the load that someone will inevitably add next year — closes the chain. Then the arithmetic, all in one line:
Required capacity (Ah) = [load Wh ÷ (system voltage × conversion efficiency × battery efficiency × usable DoD fraction)] × temperature factor × contingency factor — looked up at your discharge rate.
A worked sketch, with assumed round numbers: a 48 V site load of 2,400 W for 8 hours is 19,200 Wh. At 85% conversion efficiency and 85% battery efficiency it becomes ~26,600 Wh drawn from the bank. Across 48 V that is ~554 Ah; at a chosen 80% usable window, ~692 Ah; a cold-room factor and a 5% contingency carry it toward the 750–800 Ah class at the 8-hour rate — and the final selection is then checked against the cell’s own rate table and end-voltage curve. The assumptions are illustrative; the METHOD is the product.
Why we size against the duty, not the enquiry
Half the enquiries we receive name a capacity; the good conversations name a duty. When a buyer tells us “600 Ah” we ask what the 600 Ah must DO — because if the number came from a failing bank that was itself undersized, copying it copies the failure. Send us the five answers this article started with — loads, hours, frequency, temperature, and the voltage window your equipment tolerates — and the sizing comes back as an engineering document: capacity at your rate, cell count, and the reasoning shown. That last part is the point. A sizing you can audit is a sizing you can trust.