Where does hydrogen make sense? From renewables to electrolysis, storage and use

Make hydrogen from a wind farm’s spare power, store it, deliver it, and sort the uses worth it

Analyst lesson A16 of Energy Systems, Step by Step: follow 1 MWh of wind through an electrolyzer and back, size an electrolyzer on a wind farm’s real 2025 hours with part-load limits, compare the cost per kilogram with the owner’s net cost, size a tank for steady delivery, apply hourly and annual matching rules, and sort uses of hydrogen against their alternatives.
  • Analyst · L2
  • 30 min guided · about 30 min practice
  • After lesson 9 and lesson 7 and A8

A 100 MW wind farm with a 50 MW grid connection, every hour of 2025, and a 30 MW electrolyzer

Which power it may use Average utilization Hours running Cost per kg (partial) CO₂ of grid hours, per kg
Spare wind only 13% 2,370 $5.65 0.0 kg
Wind matched hour by hour 72% 8,555 $2.43 0.0 kg
Grid power matched over the year 100% 8,760 $2.23 5.8 kg

Average utilization is the year's output ÷ (30 MW × 8,760 hours). On spare wind the unit runs in 2,370 hours but at full size in only 205.

1 MWh of power makes 17.5 kg of hydrogen, which holds 0.58 MWh; a fuel cell turns that back into 0.35 MWh. Cost per kg counts capital and power only.

Eight uses of hydrogen, sorted by their alternative today

Use Where it sits Why
Ammonia for fertilizer Hydrogen is the feedstock Ammonia is made from hydrogen: about 60% of industry's 55 Mt of hydrogen in 2024 (IEA).
Oil refining Hydrogen is the feedstock Refineries already use over 43 Mt a year (IEA, 2024): the choice is how it is made.
Steel from iron ore (direct reduction) Hydrogen is the feedstock Hydrogen can replace the gas that strips oxygen from the ore; about 10% of industry's hydrogen today (IEA).
Fuel for ocean shipping Both routes are tried Long voyages need dense fuels; ammonia and methanol made from hydrogen compete with others.
Power storage for weeks Both routes are tried Salt caverns hold weeks of energy cheaply, but only about 31% comes back (A8).
Long-haul trucks Both routes are tried Fuel cells and batteries are both tried; a battery truck uses far less electricity per mile.
Passenger cars Electricity does it with far less Electric car: 100 miles on 25–40 kWh. Fuel cell car: about 60 miles per kg, made with 57 kWh: about 1 mile per kWh.
Heating homes Electricity does it with far less A heat pump gives up to three times the heat of the power it uses; hydrogen loses about 40% before any heat is made.

With JavaScript on, you size the electrolyzer, set its part-load limit, move the hydrogen price, size a tank and sort the uses yourself.

Illustrative farm. 100 MW of wind on the Southwest Power Pool's real 2025 hours (EIA-930), scaled to a 33.8% capacity factor (EIA, U.S. wind fleet, 2025); its grid connection takes 50 MW.

Costs, one price year. Electrolyzer $2,000 per kW, 57 kWh per kg and power at $30 per MWh, in 2022 dollars from DOE's Record 24005 (May 2024); 4% real over 30 years.

Partial cost. Upkeep and stack replacement are left out; DOE's $5–7 per kg counts them, at 50–75% utilization and its own financing. Compression, storage and delivery are outside both.

About 30 guided minutes in seven steps.

What you just did, as diagrams

Energy balance
1 MWh makes 17.5 kg; it holds 0.58 MWh and gives back 0.35 MWh through a fuel cell.
Utilization
On spare wind a 20 MW electrolyzer's average utilization is 18%: it runs in 2,408 hours, at full size in 662. Each kg carries much capital: $4.24 per kg.
Part-load limit
At 20 MW an alkaline unit (15% minimum) loses 420 MWh of spare wind a year below its minimum; a PEM unit (5%) loses 47 MWh.
Cost per kg and the system
At $4 per kg the cheapest kg comes at 5 MW, the owner's best size at 10 MW.
Storage
Steady delivery of a 20 MW unit's 2025 output needs at least 88 t of working storage, 59 days of delivery, with 44 t in the tank on 1 January and no losses.
Matching rules
A rule sets what may be claimed; what is emitted depends on what the grid does in the hours bought.

Slide the hydrogen price

  1. A higher hydrogen price makes each kilogram worth more, so the optimizer builds a larger electrolyzer to capture more surplus wind.
  2. The extra capacity runs fewer full-load hours, so the cost per kilogram, the levelized cost of hydrogen, goes up.
  3. Yet the added hydrogen revenue lowers the net cost of the whole system.
  4. The cost per kilogram and the system's cost move in opposite directions.

A separate example with stylized numbers, from the course author's open explainer: not the study's data or results.

It has its own wind shape and costs, and reads out the whole system's net cost, so its sizes and dollars differ from this lesson's farm.

Go deeper · AnalystColours, claims and matching rules
  • Colours are names, not measurements. Grey (natural gas), blue (gas with capture) and green (renewable electrolysis) are common labels; the IEA says it “does not use colours” and counts low-emissions hydrogen instead.
  • What rules measure. DOE’s clean-hydrogen target is at most 4.0 kg CO₂e per kg, well-to-gate.
  • DOE’s example. A gas reformer with about 95% carbon capture could reach about 4.0, if it uses the average U.S. grid mix and upstream methane leaks stay at or below 1%.
  • Three tests of the power. The U.S. clean-hydrogen credit’s final rules (January 2025) ask for deliverability (the same region), temporal matching (the same year or the same hour) and incrementality (new clean power).
  • Matching is not accounting. Matching sets which power may count. Separately, from 2030 DOE’s 45VH2-GREET manual (December 2025) lets a producer opt to count its electricity’s emissions hour by hour, if its annual result is at most 4.0 kg.
  • A claim is not a cause. Matching sets what a buyer may claim. Whether emissions fall depends on whether the clean power is new (additionality) and deliverable: the accounting questions of A10.
Go deeper · ModelerFrom rules to an optimization model
  • Here each hour follows a rule: run at the spare wind, or stop below the minimum load. The size is chosen by trying sizes.
  • In a model the solver chooses the size, the tank and every hour’s operation at once.
  • The tank’s balance: \(L_t = L_{t-1} + h_t - q_t\).
  • Its symbols: \(L_t\) is the hydrogen in the tank at the end of hour \(t\), \(h_t\) the hour’s production and \(q_t\) its delivery, all in kg.
  • A minimum load is a yes-or-no choice per hour (on or off), so an exact model is a mixed-integer program; a linear program can only force the unit on or ignore the limit.
  • Next. M4 builds such a plant model with a battery, a hydrogen tank and a heat store.
Go deeperWhat this lesson leaves out
  • Upkeep and stack replacement, which DOE's $5–7 per kg production cost counts.
  • Compression, storage beyond the tank, pipelines and trucks, which DOE's production cost leaves out too.
  • Credit rules and their values, which change from year to year.
  • Whether the wind farm would have been built anyway (additionality), and where its power can be delivered (deliverability).
  • Electrochemistry, stack wear and safety engineering: devices enter only as efficiencies and load ranges.
  • Water: electrolysis consumes water too.

Check yourself

Myth or fact? “A bigger electrolyzer on the same wind farm makes each kilogram cheaper.”

  • Myth
  • Fact
Answer

Myth, on spare wind. From 10 to 30 MW its average utilization falls from 23% to 13%, and the cost per kg rises from $3.25 to $5.65.

Myth or fact? “If an electrolyzer's power is matched with wind over the year, its hydrogen is made without emissions.”

  • Myth
  • Fact
Answer

Myth. Annual matching lets a 30 MW unit run from grid power in low-wind hours; if gas makes that power, it carries about 5.8 kg of CO₂ per kg, above DOE's 4.0 target.

Myth or fact? “Delivering wind-made hydrogen at a steady rate needs storage for weeks, not hours.”

  • Myth
  • Fact
Answer

Fact, for this farm in 2025. A 20 MW unit's steady delivery needs at least 88 t of working storage, 59 days of its output: windy and calm seasons take turns.

That is with 44 t in the tank on 1 January and no losses; another weather year needs its own check.

An electrolyzer uses 57 kWh per kg. How much hydrogen does 10 MWh make?

  • 570 kg
  • about 175 kg
  • about 17.5 kg
  • 5.7 kg
Answer

About 175 kg: 10,000 kWh ÷ 57 kWh per kg = 175.4. 570 multiplies instead of dividing; 17.5 kg is what 1 MWh makes; 5.7 divides 57 by 10.

Numbers, sources and what’s next

Numbers to remember

Number What it means Source
almost 100 Mt Global hydrogen demand in 2024; low-emissions production was less than 1% of global production. IEA, Global Hydrogen Review 2025, executive summary · published 12 September 2025, revised October 2025; data for 2024
$5 to $7 per kg DOE's modelled production cost of hydrogen from PEM electrolysis on renewable power, without subsidies, in 2022 dollars: upkeep and stack replacement in, compression, storage and delivery out. DOE Hydrogen Program Record 24005 · 20 May 2024
4.0 kg CO₂e per kg DOE's target for clean hydrogen: well-to-gate life-cycle emissions per kg of hydrogen. DOE, Clean Hydrogen Production Standard Guidance · June 2023
33 kWh per kg The energy in 1 kg of hydrogen, at its lower heating value. DOE Hydrogen Program Plan, p. 24 · December 2024

Next

M4 · Storage and coupled carriers: modelling power, hydrogen and heat together

In the real world

When gas-fired backup power became much more expensive, the model built fuel cells and turned stored green hydrogen back into electricity.

Industry then drew more hydrogen from gas reformers, and the round trip from power to hydrogen and back lost energy, so total emissions rose although gas-plant emissions fell.

Inputs reflect 2022–2023 conditions; post-2023 federal hydrogen credit rules are not modeled.

Case file CF3: where hydrogen should go when nobody knows how much wind will be built.

Among the cases studied, concentrated solar power with thermal storage running a high-temperature electrolyzer at variable power gave the lowest levelized cost of hydrogen; constant-power operation was the most expensive.

Approved excerpt of: Li et al. (2023), EEM 2023.

Yardstick here: the cost per kilogram across designs. Holds for: one West Texas site, cost projections from NREL's Annual Technology Baseline, a conference-length study.

Case file CF2: when the cost per kilogram and the cost of the system disagree.

Sources and further reading

Spotted an error, or stuck? Write to hlli@ksu.edu. Corrections are made in place and dated here.

Text licensed CC BY 4.0, code MIT: teachers and students may reuse and adapt it, with credit to Honglin Li.