An LCOS of EUR 0.195 per kWh cycled reads like a property of the equipment. It is not. It is the output of six assumptions, and the largest of them is not the price of the battery. It is how many times a year the site will actually cycle it, which is a fact about the site rather than about the product.
The inputs a ten-year LCOS needs
Worked example. A containerized 500 kW / 1,000 kWh system at EUR 420,000 installed, operated at 90 percent depth of discharge, with fixed maintenance of EUR 6,300 in year one escalating at 2 percent a year, one augmentation of EUR 63,000 in year 8, and a real discount rate of 6 percent over a ten-year horizon.
The formula is total discounted cost divided by energy delivered to the load:
- LCOS = present value of all costs / total kWh delivered to the load over the horizon
- Energy per cycle = nameplate kWh x depth of discharge x round-trip efficiency
- Present value of maintenance = sum of annual cost x 1.02^(y-1) / 1.06^y for y = 1 to 10
The arithmetic
Worked example. The assumption set used here differs from the one behind the published and the published , which assume 90 percent round-trip efficiency and 365 cycles a year. Neither set is wrong. They describe different sites, and the 13 percent gap between the last two rows is the point of this article.
What the result means
LCOS is the price per unit of energy the system actually moves, which means it only carries meaning next to the value of whatever it displaces. If the displaced value is a per-kilowatt-hour tariff spread of EUR 0.16, an LCOS of 0.195 does not pay for itself on arbitrage alone.
That comparison fails for most commercial sites, because their value is not per kilowatt-hour at all. Demand-charge reduction is paid per kilowatt of avoided peak, and the installed systems that work hardest on it are often small: a and , both completed in May 2026, earn their return from demand reduction and sold services rather than from buying low and selling high.
Ruibit Energy delivers container systems into industrial-park and factory duty cycles, and the line that decides this arithmetic is not the installed cost but the cycles per year the site will genuinely run.
When the number stops holding
Three sensitivities matter more than the rest, and they are all numeric.
- Cycles fall from 330 to 250 a year: energy delivered falls to 1,980,000 kWh and LCOS rises to EUR 0.257, which is 32 percent higher
- Round-trip efficiency falls from 88 to 85 percent: energy per cycle falls to 765 kWh and LCOS rises to EUR 0.202
- Both sets of published assumptions are used: EUR 0.172 against EUR 0.195 for the same hardware
If the site's own 15-minute data shows fewer than 250 cycles a year, the number to carry into the investment case sits above EUR 0.257, not at the EUR 0.195 on the quotation. If augmentation is financed rather than paid from cash flow, the cost of that capital belongs in the numerator and moves the figure again.
The boundary is narrower than it looks. The numerator is a present value and the denominator is not, so a fully discounted LCOS comes out higher. The figure counts only cycled energy, which means it cannot be set against a demand-charge saving, a backup-hours requirement, or a capacity market payment without converting them first. Augmentation beyond the first event, the financing cost of that augmentation, residual value and local incentives are all outside the example.
What to ask before accepting an LCOS figure
Ask how many cycles a year the figure assumes, and whether that number comes from the site's own 15-minute data or from a datasheet maximum. An answer of 365, or one cycle a day, describes a dispatch pattern most industrial sites never run. The states a cycle life of at least 8,000, which spread over a ten-year horizon is 800 cycles a year, so cell endurance is rarely the binding constraint. The duty cycle the site can actually create is the binding constraint.
Ask whether the denominator is energy delivered to the load or energy out of the power conversion system. Those differ by the round-trip losses, and the difference is around 12 percent on this example.
Ask whether augmentation is included and whether the number is discounted. A ten-year claim that excludes the mid-life augmentation is not a ten-year cost.
Once the purchase order is signed, three of these stop being negotiable. The warranty throughput limit sets a ceiling on cycles regardless of what the site wants to run. The augmentation obligation, or its absence, is a contract term rather than an operating decision. And the metering boundary determines which kWh the warranty and the savings claim are measured against. Under a 250-cycle duty cycle this system costs EUR 0.257 per kWh cycled, which is above the tariff spread that was supposed to pay for it. That is the condition under which the number stops arguing in favour of the project and starts arguing against it.
FAQs
1. What is the levelized cost of storage?
The cost per kilowatt-hour actually cycled, calculated as the present value of all costs over the project horizon divided by the total energy delivered to the load. It is a cost per unit of energy moved, not a purchase price.
2. How is LCOS calculated for a commercial BESS?
Total discounted cost divided by energy delivered. Energy per cycle is nameplate capacity multiplied by depth of discharge and round-trip efficiency, so 1,000 kWh at 90 percent depth of discharge and 88 percent round-trip efficiency delivers 792 kWh per cycle.
3. Why do two LCOS figures for the same system differ?
Because the denominator is a duty cycle, and the assumption set is chosen by whoever presents the number. The same hardware returns EUR 0.195 per kWh cycled at 330 cycles a year, and EUR 0.172 at 365 cycles a year with 90 percent round-trip efficiency.
4. Can LCOS be compared with my electricity tariff?
Only if the displaced value is also per kilowatt-hour. A tariff spread of EUR 0.16 does not cover an LCOS of EUR 0.195. Demand-charge reduction is paid per kilowatt of avoided peak, so it cannot be set against a per-kWh figure without conversion.
5. What does LCOS leave out?
Augmentation beyond the first event, the financing cost of that augmentation, residual value, local incentives, and any value that is not per kilowatt-hour, such as backup hours. The numerator is discounted while the denominator is not, so a fully discounted LCOS comes out higher.
