Your LCOH is wrong, because your stack doesn't run at BOL efficiency for 20 years.
Almost every published levelized-cost-of-hydrogen figure assumes constant beginning-of-life efficiency for the whole project. Real stacks degrade. This calculator models the voltage rise, the operating mode it actually matters under, and the stack replacement most spreadsheets forget — then shows you the gap in dollars per kilogram.
Plant
Power fixed (grid cap / PPA) — production falls as the stack degrades.
Degradation
Costs & finance
Adjust inputs and click Compute LCOH.
This is one plant, one set of assumptions, computed once. HYDRA OS recomputes degradation-adjusted LCOH continuously from live stack data — no re-pasting numbers every quarter.
These are the values this page loads with. Open the tool and press Compute LCOH and you will get exactly these numbers back — they are produced by the same engine the page runs, so the example is a reproducible check, not an illustration.
Inputs — 10 MW PEM plant
Rated power
10 MW
Capacity factor
55 %
BOL specific energy
52 kWh/kg, system AC
Degradation
3 µV/h (cell basis)
Operating mode
Constant power
Replacement trigger
10 % voltage rise
Stack share of capex
40 %
Capex
1,500 $/kW
WACC
8 %
Electricity
50 $/MWh
Fixed O&M
2 % of capex/yr
Water
12 L/kg at $3/m³
Project life
20 years
Results
Naive LCOH
$4.61 /kg
Degradation-adjusted
$5.06 /kg
Difference
+$0.45 /kg (+9.8%)
Stack replacements
1 (year 12)
Specific energy
52.22 → 55.56 kWh/kg, yr 1 → yr 20
Lifetime output
17.76 million kg (naive: 18.53 M)
Levelized cost PV
$44.3 million
Discounted output
8.75 million kg
A 3 µV/h degradation rate — unremarkable for a PEM stack under a variable load — moves this project's levelized cost from $4.61/kg to $5.06/kg, and buys one stack replacement in year 12 that the constant-efficiency model never sees at all.
Note on the degradation input: 3 µV/h is entered on the conventional cell-voltage basis, while the BOL specific energy field is on a system/AC basis. The tool scales between the two using the same rectifier and balance-of-plant factors as the Spec Normalizer (÷0.96, ×1.12), so the effective rate reaching the model is 3.50 µV/h. Mixing the two bases without that correction understates the damage by roughly a whole stack replacement over 20 years.
Method & sources
Degradation is modeled as a uniform rise in cell voltage over operating hours — the standard simplification behind every published µV/h or %/1000h figure. That voltage rise converts to specific energy through the same relationship used across all five tools: kWh/kg = 26.59 × V_cell ÷ Faradaic efficiency.
The mode that most LCOH tools get wrong
Constant current: production/hour is fixed (Faraday's law) → kWh/kg rises, kg/year does not
Constant power: energy/hour is fixed → kWh/kg rises → current falls → kg/year falls
Most real projects — anything coupled to a capped grid connection, a PPA, or curtailed renewable supply — operate closer to constant power. Most published LCOH models implicitly assume constant current (or ignore degradation entirely), because it's the easier case to compute. That understates how much degradation actually costs.
The levelized cost formula
Σ [CAPEX + FOM + Energy + Water + VOM + Stack Replacement − Subsidy] / (1+r)^t
LCOH = ─────────────────────────────────────────────────────────────────────────────────
Σ [Production_t / (1+r)^t]
The production stream is discounted at the same rate as the cost stream — the LCOE convention. Skipping this (discounting costs but not production) understates LCOH whenever production is front-loaded relative to costs, which is exactly what happens once degradation and a mid-life stack replacement are modeled honestly: early years produce more per dollar spent than late years do.
Stack replacement
A replacement is scheduled whenever cumulative degradation pushes cell voltage past your trigger threshold (default +10%, a common warranty/EOL convention), at a cost equal to your stated share of total CAPEX (default 40%). Note this is a replacement-specific share, not the initial stack's share of a new system: a stack's share of a brand-new system's CAPEX varies widely by technology and scope (roughly 19–60% depending on what's bundled in), but replacing just the stack module — no fresh BOP, civil works, or power electronics — is consistently narrower across published PEM/AWE cost studies, in the 30–40% range. 40% sits at the top of that band on purpose (a higher assumed replacement cost is the conservative direction for LCOH). The degradation clock resets after each replacement; a long project life with a high degradation rate and low trigger threshold can schedule more than one.
The "naive" comparison
The naive figure shown alongside your result uses your BOL specific energy held constant for the entire project life, with no stack replacement scheduled — because that is what a genuinely naive LCOH calculation does, not just a simplification of one. The gap between the two numbers is not a modeling nuance; it's the cost of an assumption.
Default coefficient sources
Every default below is shown in the form and freely overridable — these are starting points, not your project's numbers:
Fixed O&M, 2%/yr of CAPEX — published techno-economic studies for electrolyzer systems commonly report a 1–3%/yr range, with 2.5%/yr a frequently-used baseline figure; 2% sits inside that range.
Water price, $3/m³ — this is industrial/demineralized process water delivered to electrolyzer feed spec, not raw municipal tap water. Raw water pricing alone varies enormously by region (roughly $0.05–3.50/m³ across US–EU, OECD average on the order of $1.20/m³); treatment to feed quality (deionization, degassing) adds real cost on top, which is why this default sits toward the higher end of the raw-water range rather than at the regional average.
Water use, 12 L/kg — the stoichiometric minimum from the 2H₂O → 2H₂ + O₂ mass balance is ≈9 L/kg; real industrial PEM systems (including purification, cooling, and rejection losses) are commonly reported in the ≈9–25 L/kg range. 12 sits near the stoichiometric end.
WACC, 8% — a commonly-used illustrative discount rate in published utility-scale renewable/electrolysis LCOH studies (typical cited range roughly 6–10%, depending on project risk and geography). This is not a substitute for your own project's actual cost of capital.
What this tool deliberately does not do
It uses a flat electricity price, not an hourly profile — a hookup for hourly renewable/spot-price profiles is a natural next iteration but isn't in this version. A flat price is a reasonable approximation when capacity factor is set correctly, but it can't capture price-degradation correlation (e.g. curtailment-driven low prices coinciding with low-load, low-degradation operation).
It doesn't model inflation separately — all inputs are treated as real (inflation-adjusted) terms, consistent with using a real WACC.
It assumes uniform (not mechanism-resolved) degradation — the same simplification the underlying µV/h figure already makes.