When Does Long-Duration Storage Actually Beat Lithium-Ion?

Long-duration storage chemistries get pitched as "superior" to lithium-ion because they last longer and skip Chinese supply chains. Both true. Neither is the point. What matters: over its service life, what does it cost per unit of energy it actually delivers?

As grids add more variable renewables, long-duration storage (LDES) keeps getting flagged as essential for multi-day reliability. Fair — but judging it on installed cost per kWh alone skips the constraints that decide whether that capacity ever gets used: round-trip efficiency, balance-of-plant scaling, energy density, and charging latency.

A metric built for lifetime throughput

We use a version of levelized cost we call LCUS — Levelized Cost of Using Storage: total lifetime cost, discounted, over total energy delivered, discounted.

Amortization is where long-duration tech struggles. A 4-hour lithium-ion asset cycling daily hits ~7,300 cycles over 20 years; a very-long-duration asset cycling four times a year hits 80 — a 90x gap in energy delivered per installed kWh. So the long-duration asset needs an installed cost nearly two orders of magnitude lower to match lithium-ion, if asked to do the same job.

But at a genuine four-cycles-a-year requirement, both technologies carry the same low-utilization handicap. Then it comes down to upfront cost and the details below — not the better slogan.

Where the real numbers land

Uninstalled LDES media is often quoted at $15–35/kWh — compelling until balance-of-plant reality (plumbing, hydrogen management, DC bus complexity, lower density) sets in. Offsetting that: a 100-hour discharge rate needs far less power-conversion hardware than 4-hour lithium-ion — in one model, EPC installation ran ~$11/kWh for long-duration versus ~$16/kWh for lithium-ion.

Efficiency is the expensive part. RTE near 45% (versus ~90% for lithium-ion) means more electricity per kWh delivered and a much longer recharge: cycle time runs discharge period × (1 + 1/RTE), so a 100-hour asset at 45% RTE takes 320 hours per full cycle — a hard ceiling of 27 cycles a year, beyond which the LCUS curve flattens into a floor.

The crossover point

Modeled comparison: iron-oxide LDES vs. 4-hour lithium-ion, using representative CAPEX, RTE, and service-cost assumptions.

In this worked example — iron-oxide at 45% RTE, $35/kWh versus lithium-ion at 90% RTE, $130/kWh — the crossover lands around 60 cycles a year. Below that, long-duration is cheaper to operate; past it, lithium-ion pulls ahead and keeps widening the gap. Better RTE, cheaper charging energy, or a higher power-to-energy ratio shifts the crossover right, but doesn't erase it.

What this means for technology selection

A handful of cycles a year — rare outages, seasonal balancing — and the lower-upfront-cost option can be the right economic call despite weaker RTE. Frequent grid services, and the efficient, high-cycling asset wins despite a higher price tag. Neither technology is "better." The duty cycle decides — run it through LCUS before comparing capex.

Want the full model?

The complete technical whitepaper includes the LCUS derivation, the sensitivity analysis on RTE and charging cost, and the full crossover chart set.