Quick Answer: For a 10-year commercial BESS project, size the initial capacity so that the end-of-warranty capacity (typically 70% of nameplate) still covers the site's daily throughput — not today's throughput. The formula: required initial usable capacity = daily energy throughput ÷ EOL factor ÷ round-trip efficiency. A site cycling 1.8 MWh/day through a 94%-efficient system needs about 2.0 MWh usable at year 1 to still deliver 1.8 MWh in year 10 at 70% EOL. Adding a 10–20% operating margin on top of that is good practice; oversizing beyond ~30% usually wastes capital that never cycles.
Why oversizing is a math problem, not a preference
Three numbers determine the margin:
- Daily energy throughput — how many kWh the battery must move per day (peak shaving + arbitrage + backup combined)
- End-of-warranty (EOL) factor — the capacity the supplier guarantees at year 10, usually 70–80%
- Round-trip efficiency — the loss inside the battery, typically 92–95%
The error most project owners make: they size for today's throughput, add a round "20% margin," and never check whether the last year of the contract still works.
The worked example
A commercial site needs the BESS to move 1.8 MWh per day — 900 kWh in the morning peak, 900 kWh in the evening peak, at 90% DoD per event.
Step 1 — usable energy at year 1: 1.8 MWh ÷ 0.94 RTE = 1.91 MWh usable per day (battery-side energy)
Step 2 — EOL adjustment: 1.91 ÷ 0.70 EOL = 2.74 MWh usable at year 1
Step 3 — convert to nameplate (90% DoD): 2.74 ÷ 0.90 = 3.04 MWh nameplate
So a site that "only needs" 1.8 MWh/day today should order roughly a 3 MWh system — 67% more than the naive 1.8 MWh answer. That is the real oversizing requirement, and it comes from the EOL guarantee, not from optimism.
The three checks that set the final number
Run all three; the largest result wins.
Check 1 — EOL capacity check (above). Initial usable ≥ daily throughput ÷ EOL ÷ RTE.
Check 2 — Cycle-life check. Divide the 10-year throughput by the warranty cycle count.
- Yearly throughput: 1.8 MWh/day × 365 = 657 MWh
- 10 years: 6,570 MWh
- If the battery warranty allows 6,000 cycles at 90% DoD on 3.04 MWh nameplate → 6,000 × 3.04 × 0.90 = 16,400 MWh lifetime. Plenty.
- But a 2 MWh nameplate would give 6,000 × 2 × 0.90 = 10,800 MWh — still above 6,570, yet the EOL check above already failed. So EOL is the binding constraint here.
Check 3 — Peak-power check. Initial PCS power must cover the largest single event: peak demand to shave, or backup critical load, whichever is larger. Power and energy are different constraints; an energy-correct system can still fail a 1-hour 800 kW event if the PCS is undersized.
What a supplier's capacity guarantee should say
Buyers should put the margin assumptions into the contract:
- EOL capacity — "≥ 70% of nameplate at year 10" or "≥ 70% at 6,000 cycles, whichever comes first"
- Throughput cap — the cumulative MWh the warranty covers. Beyond the cap, the capacity guarantee stops applying even if calendar time has not expired.
- Test conditions — the EOL test is at 25°C, 0.5C, specific DoD. Hotter sites age faster; a site at 35°C ambient should use 65% EOL in the arithmetic, not 70%.
The oversizing sweet spot
From the worked example and typical EU commercial tariffs:
| Margin over year-1 usable need | Effect |
|---|---|
| 0% (exact EOL math) | Passes year 10 only if every assumption holds — no room for hotter summers or deeper events |
| 10–20% | Covers measurement error, temperature, and a bad year of cycling; typical good practice |
| 25–30% | Adds headroom for load growth; check the payback still works |
| >30% | Capital sits idle; cycles stay shallow, but the extra MWh rarely earns its cost |
Rule of thumb: size the EOL math, then add 10–20% operating margin, and stop. More than that is usually a sales argument, not an engineering one.
The "measured vs published" discipline
The EOL factor on the datasheet is lab data. Real sites run warmer, cycle irregularly, and hit events deeper than assumed. Two practical corrections:
- Use the supplier's tested degradation curve if published, not the marketing line
- Ask for the throughput cap in writing — a generous EOL number with a small throughput cap is a trap
What to send the supplier
To get a correct oversizing recommendation, provide:
- 12 months of 15-minute interval data (peak size and duration per event)
- The daily throughput you intend (peak shaving + arbitrage cycles)
- Site ambient temperature range
- The tariff structure (how many events per day the EMS will run)
With those, a supplier can compute the EOL, cycle, and peak checks and quote the smallest system that still works in year 10. Ruibit runs exactly this three-check sizing for Dawnice commercial cabinets — EOL capacity, cycle throughput, and peak power — and returns the recommended nameplate with the margin stated, before the PO.
FAQs
Q: Is oversizing the same as buying a bigger battery? Yes and no. Oversizing means buying more energy capacity than today's need, specifically to cover EOL degradation and operating margin. It is the calculated difference, not a round 20%.
Q: What EOL factor should I use for the calculation? Use the supplier's guaranteed EOL (typically 70–80%). In hot climates, drop it by 5 percentage points (e.g., 70% → 65%) to account for accelerated calendar aging.
Q: Does a larger battery last longer in cycle life? Yes, because it cycles shallower for the same throughput. But the extra capital must earn a return — check the payback before adding margin beyond ~20%.
Q: What is the throughput cap and why does it matter? The cumulative MWh the warranty covers. Even a battery with a 10-year calendar guarantee stops being capacity-guaranteed once the throughput cap is reached. Compare cap × capacity against your 10-year throughput.
Q: Can I expand the BESS later instead of oversizing now? Yes, if the system supports parallel expansion and the modules stay available. This lowers first cost but depends on product continuity — confirm module availability and compatibility in writing before relying on it.
