How it runs — the feasibility run settings

The feasibility design panel’s How it runs section holds five choices. The ⓘ beside each opens one section of this page — its first paragraph and each option’s first sentence, then the rest with Compare in full: this page is where that text lives. The options are listed in the order the panel shows them. How the two engines work is in How feasibility works.

Engine

Both engines run the same plant on the same hourly year. They differ in what they know: rules decide hour by hour with no view of the weather ahead; the solver plans the whole year at once.

The battery and H₂-buffer rules and the promised output apply to the rule-based engine only. The solver works out the best storage schedule by itself, so imposing a rule on it could only make it worse.

Rule-based

Fixed rules, hour by hour, with no view of the weather ahead.

1. Sorts the hour into one of three cases: not enough power, enough power, or surplus power.
2. Sets the electrolyser’s load from your battery strategy, within its ramp limit.
3. Feeds the synthesis units from the hydrogen just made plus the tank, following your H₂ buffer policy.
4. Carries the battery charge and tank level into the next hour.

Trade-off Runs in well under a second, and behaves the way a real operator would. Slightly optimistic on tonnes: units kept at their minimum rate on backup supply still count that output, and those same hours are reported as shortfall.

Choose it This is the number you defend to a lender, and the one to use while iterating.

Solver (LP)

Plans the whole year with the weather known — the best case.

1. Builds one linear program covering all 8,760 hours (PyPSA) and solves it with HiGHS.
2. Obeys exactly the same operating limits as rule-based: minimum loads, ramps, standby draw, battery and tank sizes.
3. It may charge today for a shortage three weeks away — information no real operator has.

Trade-off The best schedule physically available, so it reaches the fewest hours below minimum rate. Takes roughly 10–15 seconds per year.

Choose it Use it as the ceiling you measure against, not as an operating plan. The distance from rule-based is what better control or automation could be worth here.

Battery

Wind and sun arrive when the weather provides them, not when the electrolyser wants them. Each hour this rule sets what load the electrolyser aims for, and whether the battery may discharge to help it get there. That second part is the real difference: Follow supply never discharges to raise output; the other three do, each in a different pattern.

Follow supply

Runs on the wind and sun as they arrive; the battery only covers must-run loads.

1. The electrolyser aims for its full rate, but only on the renewable power actually arriving that hour.
2. The battery does not discharge to push the electrolyser harder.
3. It is used only to keep the synthesis units and standby load running.
4. Surplus power charges the battery; once it is full, the rest is curtailed.

Trade-off On one measured year of a real site this was the most reliable by a wide margin — 43 hours short of power against 4,315 for Push output — and the easiest on the battery, cycling 33 GWh against 93. It gives up only 2.6% of hydrogen to do it.

Choose it The sensible default. Pick something else only if you have measured that the extra output is worth the reliability it costs.

Push output

The battery also pushes the electrolyser toward full rate when power is short.

1. The electrolyser aims for its full rate every hour.
2. When renewables fall short, the battery discharges to close the gap — limited by its MW rating and the charge it holds.
3. Surplus power recharges it, so it is emptied and refilled often.
4. Because the electrolyser is served before the synthesis units, the battery it spends is no longer there for the must-run load.

Trade-off The most hydrogen of the four, but the margin is small and the cost is not: on one measured year of a real site it bought 2.6% more hydrogen and paid 4,272 extra hours of power shortfall — roughly 100 times less reliable. It also cycles the battery ~3x harder.

Choose it Only when output is genuinely worth more than firmness, and the battery is sized generously enough that it is not empty when a long lull starts.

Steady rate

One steady electrolyser rate all year; the stores absorb the weather.

1. One setpoint is worked out in advance: the average hydrogen rate the renewables can sustain, capped by what the synthesis units can take.
2. Above the setpoint, surplus charges the battery; below it, the battery discharges to hold the rate.
3. Peaks are deliberately not chased, so the electrolyser barely ramps.

Trade-off Gentle on the ELECTROLYSER — about half the up-and-down cycling, which is what wears a stack — but hardest on the BATTERY: on one measured year of a real site it cycled 116 GWh, the most of all four, against 33 for Follow supply. It was also the weakest overall there: 17% less hydrogen, curtailment 37% against 22%, and still 2,401 hours short of power.

Choose it When electrolyser stack life is the binding concern and you accept paying for it in output, curtailment and battery wear. Check the battery throughput before choosing it.

Store, then run hard

The battery charges fully, then spends it in one sustained push.

1. Works like Push output, except the boost is switched on and off by how full the battery is.
2. It charges until it reaches 80% full, then starts boosting the electrolyser.
3. It keeps boosting until it drops to 30% full, then stops and recharges.

Trade-off Fewer, deeper battery cycles instead of many shallow ones, and production arriving in long runs. On one measured year of a real site it sat between the other boosting options: 1,497 hours short of power against 43 for Follow supply and 4,315 for Push output.

Choose it When you want long uninterrupted runs downstream, or to avoid constant shallow cycling, and can accept the reliability cost of boosting at all.

H₂ buffer

The electrolyser’s output rises and falls with the weather, but the synthesis units have to keep running above their minimum rate or shut down. This rule sets how much hydrogen the units draw each hour: everything available now, or part of it held back for later.

Why the choice matters Every hour a unit cannot hold its minimum rate counts as a shortfall — the hours short of hydrogen on the Reliability tab. Using hydrogen as soon as it is made gives more product now; holding some back gives fewer shortfall hours later. Neither is right in general, which is why it is yours to set rather than ours to assume.

Where the numbers come from The steady rates used by “Steady rate” and “Keep a reserve” are each worked out once, from the average rate the renewables can sustain across the year. They are design settings, not hour-by-hour foresight — the rule-based engine never looks ahead. The reserve levels (50% and 80% of the tank) are assumptions you can edit in the assumption book, as h2_buffer_band_lower_pct and h2_buffer_band_upper_pct.

Use freely

Units run as hard as the hydrogen allows; the tank holds almost nothing back.

1. Each hour the units take their minimum first, then as much extra hydrogen as there is.
2. One amount is always left in the tank: enough to cover one hour at minimum rate, so the tank can still bridge the start of the next quiet period rather than sitting empty.
3. Hydrogen the units cannot take is stored; once the tank is full, surplus power is curtailed.

Trade-off The most product. Because the tank is kept nearly empty, a long low-wind period starts causing shortfall sooner than it would under Keep a reserve.

Choose it When hydrogen is plentiful compared with the synthesis units, or product tonnes are what you are judged on.

Keep a reserve

Units hold a steady rate; the tank is kept as a reserve.

1. A steady synthesis rate is worked out in advance: the average hydrogen rate the renewables can sustain.
2. While the tank sits between 50% and 80% full, the units simply hold that rate.
3. Below 50%, they run under the rate so the tank refills and the reserve is rebuilt.
4. Above 80%, they run over the rate to draw the tank back down.

Trade-off Fewer hours below minimum rate, because the reserve covers quiet periods. Slightly less product, since some hydrogen is deliberately held back and the steady rate does not chase peaks.

Choose it When shortfall hours are the problem you are trying to fix, or the plant has to look firm to an offtaker.

Promised output

Not a dispatch preference: it decides what a shortfall IS, and therefore which number on every reliability screen counts as a failure.

Firm, always on

Output is owed every hour — a unit below its minimum is a broken promise.

1. Every process unit is held at its minimum rate in all hours.
2. When there is not enough hydrogen to do that, the shortfall is recorded as an unserved hour.
3. The tonnes it would have made at that minimum are still counted, so production includes output the hydrogen could not support.

Trade-off The strictest reading. Hours short of hydrogen count against you, and reported production is slightly higher than the plant can really deliver.

Choose it A pipeline or contracted offtake that must be met hour by hour.

Annual tonnage

A yearly quantity is owed — a unit with no hydrogen simply stops.

1. A process unit short of hydrogen is shut down for that hour instead of being propped up.
2. It produces nothing and draws no standing electrical load while it is down.
3. Those hours are reported as time stood down, not as unserved hours.
4. Units are stopped from the bottom of the derivative list upward, so the first one listed keeps running longest.

Trade-off Honest tonnage and an honest power balance: hydrogen shortfalls stop counting as failures, and reported production falls to what the plant actually makes.

Choose it Merchant sales, or any offtake measured over a year. Also the right setting while you are sizing, so output is not flattered.

Horizon

How much of the plant’s life is simulated, and whether equipment ageing is applied.

Single year

One representative year, with the plant at full, as-new performance.

1. Simulates 8,760 hours — one full year.
2. No ageing is applied: the electrolyser runs at its new efficiency.

Trade-off Fastest. Shows what the plant does in a typical fresh year, so it says nothing about how output falls as equipment ages.

Choose it The default. Use it while you are still sizing and comparing designs.

Checkpoints

The same single year re-run at a ladder of plant ages (1, 5, 10, 15, 20…).

1. Each checkpoint applies that year’s ageing: electrolyser efficiency loss and resource degradation.
2. Years in between are filled in from the checkpoints either side.

Trade-off A handful of solves instead of thirty. Gives the lifetime trend and the lifetime cost, but it never sees an unusually bad individual year.

Choose it When you want the lifetime picture without paying for every year. Works with both engines.

Every year

Every year of the plant life, run one after another in order.

1. Battery charge, hydrogen in the tank and the electrolyser’s accumulated age carry across each year boundary.
2. The solver also looks three weeks past each year end, so storage is not emptied artificially in December.

Trade-off The complete year-by-year record. Seconds with the rule-based engine; several minutes with the solver, which runs a full optimisation per year.

Choose it When you need true per-year detail — replacement timing, or the worst year for reliability.