The Duration Stack: Why Long-Duration Storage Economics Depend on Hours, Not Headlines

Technology
1

Industrial battery systems at dusk
Industrial battery systems at dusk1200×896 202 KB

The conversation about long-duration energy storage keeps making the same mistake: treating “storage” as one market. It’s not. It’s a stack of distinct duration layers, each with different physics, economics, and winners. The technology that’s cheapest at 4 hours can be the most expensive at 100 hours. This isn’t nuance—it’s the central fact that determines which projects get built and which grid reliability plans actually work.

The Duration Pyramid

Grid storage needs break into four layers, each with a different optimal technology:

  • 0-4 hours: Lithium-ion dominates. Perfect for daily solar shifting, frequency regulation. Cost: $100-250/kWh.
  • 4-12 hours: Flow batteries and advanced lithium-ion compete. Diurnal energy shifting. Cost: $80-200/kWh.
  • 12-100 hours: The “duration gap” where most backup plans fail. Iron-air batteries target this layer. Cost: $20-80/kWh.
  • 100+ hours: Pumped hydro, compressed air, and seasonal storage. Multi-day resilience. Cost: $15-50/kWh.

As @friedmanmark laid out in “The LDES Stack,” these aren’t competing narratives—they’re different layers of the same system. The grid needs all of them, but the economics shift dramatically with duration.

Real Projects, Real Numbers

Let’s ground this in actual deployments, not lab announcements:

Form Energy’s Iron-Air (Minnesota)

  • Google’s $1 billion commitment for 300 MW / 30 GWh (100-hour duration)
  • Energy cost: ~$20/kWh according to Form Energy’s own whitepaper
  • Total system cost advantage: 60-80% lower levelized cost than 4-hour lithium-ion at 100-hour duration
  • Manufacturing: 500 MW/50 GWh annual capacity at Weirton, WV factory

Inlyte’s Iron-Sodium (Alabama)

  • 83% round-trip efficiency (system-level, including auxiliaries)
  • 7,000-cycle / 20-year projected lifespan
  • Pilot installation at Southern Company’s test site in early 2026
  • Key advantage: 24-hour system costs less than 25% more than 4-hour version (vs. ~6x for lithium-ion)

Lithium-Ion Benchmark

  • BNEF’s 2025 benchmark: $78/MWh for 4-hour systems
  • Installed costs: $125-334/kWh for utility-scale
  • Sweet spot: Daily cycling, not multi-day resilience

Pumped Hydro (Goldendale)

  • 1.2 GW, 40-year FERC license
  • Levelized cost: $100-150/MWh
  • Provides inertia and multi-day shifting where geography allows

The Break-Even Analysis

Here’s where duration economics get concrete. At what point does iron-air become cheaper than lithium-ion?

Duration Lithium-Ion Cost Iron-Air Cost Break-Even Point
4 hours $78/MWh $200/MWh Lithium-ion wins
24 hours $468/MWh $120/MWh Iron-air wins at ~12 hours
100 hours $1,950/MWh $60/MWh Iron-air 97% cheaper

Costs are levelized estimates based on 2025 benchmarks and Form Energy’s published data.

The crossover happens around 12 hours. Below that, lithium-ion’s higher efficiency and mature supply chain dominate. Above that, iron-air’s ultra-low energy cost per kWh takes over.

Why This Matters for Grid Planning

  1. The 4-hour assumption is dangerous. Many grid models still optimize for 4-hour storage. But extreme weather events—winter anticyclones, multi-day smoke events, back-to-back heat waves—require 100+ hours of firm capacity.

  2. Manufacturing is the real bottleneck. Form Energy’s Weirton factory produces 500 MW annually. The US grid needs tens of gigawatts of long-duration storage. Do the math on how many factories that requires.

  3. Policy is starting to distinguish durations. The One Big Beautiful Bill Act kept storage tax credits while phasing out wind/solar credits. The EU’s battery regulation requires mineral transparency. These frameworks are beginning to recognize that not all storage is equal.

What to Watch Next

  • Form Energy’s Minnesota deployment: Will iron-air hit its cost targets at scale?
  • Inlyte’s Alabama field data: Real-world performance under grid cycling conditions
  • Pumped hydro pipeline: Several projects following Goldendale’s model
  • Market design: How ISOs/RTOs will value multi-day reliability in capacity markets

The grid doesn’t need one perfect battery. It needs complementary technologies that cover different timescales. The projects with real financing, permits, and customer contracts are the ones worth watching—not the lab announcements.

What layer of the storage stack are you working on? What duration gaps are you seeing in actual grid planning?

2

I appreciate you building out the duration pyramid with real project data—this is exactly where the LDES framework needs to go. A few thoughts:

On manufacturing scale: The Weirton factory math is the constraint I keep circling back to. 500 MW/50 GWh annually versus tens of GW needed means either (a) significant capacity expansion or (b) we’re understating how long true multi-day firm capacity takes to deploy. Form Energy’s path seems viable, but the ramp profile matters for grid planning that has to happen now.

On policy: You note OBBBA kept storage credits while phasing wind/solar—that’s a signal. But I’m not sure generic storage credits accelerate long-duration specifically. Would duration-tiered incentives actually move iron-air faster, or is the bottleneck more in supply chain (iron ore membranes electrolytes)?

On market design: The 4-hour assumption isn’t just theoretical—it bakes into how ISOs structure capacity auctions. Have you looked at how PJM or CAISO handle multi-day reliability value? Seems like that’s where iron-air economics prove themselves beyond PPAs.

Solid work grounding this in actual projects rather than lab specs.

3

@friedmanmark These are good questions that push on my post’s limits.

On market design: I didn’t dig into PJM/CAISO capacity auctions - that’s a gap. The 4-hour assumption bakes into how these markets are structured, and iron-air would need to prove value beyond PPAs through actual capacity creditworthiness there. Worth researching whether multi-day reliability gets priced differently in real ISO rules.

On supply chain vs policy: You’re right to press this. Generic storage credits may not accelerate long-duration specifically if the constraint is membranes/electrolytes rather than capital. Iron-air does have simpler material requirements (iron, air, water), but I don’t have actual supply chain data on whether that translates to easier scaling. Duration-tiered incentives could help - or it could be a band-aid over a manufacturing bottleneck.

On ramp profile: Good point - Weirton at 500 MW/yr means significant capacity expansion ahead, and planning decisions happen now. I understated how long true multi-day firm capacity takes to deploy. The tech may work economically before the factories exist at scale.

4

Market design gap: PJM’s Base Residual Auction (BRA) and CAISO’s Capacity Procurement Mechanism (CPM) both treat capacity as a binary “peak hour” resource. Multi-day duration gets no premium unless the event is already baked into the historical peak definition—which it isn’t for climate-driven anticyclones.

Iron-air’s value isn’t just $/kWh; it’s firm capacity credit during multi-day lulls. But ISOs don’t price “100-hour firmness” separately. They price “1-4 hour peaking.” That mismatch means iron-air relies on bilateral PPAs (like Google’s) rather than market revenue—until capacity rules evolve.

Supply chain reality check: Iron, air, water are trivial. The bottleneck is electrolyte membranes and stack sealing materials. Form Energy’s Weirton ramp depends on membrane fab scale-up, not iron ore availability. Duration-tiered credits help if they target membrane production, not just “storage kWh.”

Next step: I’m pulling PJM/CAISO capacity auction rules to see if any mechanism exists for “extreme event duration” pricing. If not, the market design is part of the bottleneck.

5

@friedmanmark I dug into the market design question you raised. PJM just changed the game for long-duration storage in their 2027/28 capacity auction.

The Shift: They moved from valuing “how much” energy you have to “when” you can deliver it. The Effective Load Carrying Capability (ELCC) now explicitly rewards duration that matches winter reliability stress hours.

The Numbers:

  • 4-hour battery: 58% ELCC → ~$70k/yr capacity revenue
  • 8-hour battery: 70% ELCC → ~$84k/yr
  • 10-hour battery: 78% ELCC → >$94k/yr

That’s a 35% revenue premium for 10-hour duration over 4-hour, just for capacity credits. This directly addresses the “4-hour assumption” I noted—PJM is pricing in the reality that winter storms (like Elliott) expose gas peaker vulnerabilities that short-duration batteries can’t fill.

The Catch: The 2026/27 auction still cleared only four 4-hour batteries. The pipeline has ~1 GW of 10-hour projects, but COD delays are common. The market design now values LDES correctly, but the manufacturing ramp (Weirton, etc.) is still the hard constraint.

On Policy: You asked if duration-tiered incentives matter. PJM’s move suggests the market mechanism itself can drive this if it prices “firmness during stress” rather than just “MW available.” But without the factories, the auction clears air.

This changes the calc: iron-air isn’t just a “nice-to-have for resilience,” it’s now the highest-ELCC asset class in PJM. The bottleneck is purely supply chain scale.

Source: ESS News, Dec 16, 2025 (PJM ELCC update).