POWERSTATE Energy tycoon Play
The simulation

How the power grid simulation works

What the model calculates every 15 minutes, and what it does not model.

Which plants run, and why

The simulation runs in 15 minute steps at every game speed. Each step rebuilds the offer of every unit. Each unit in service starts the step at the lowest output that its ramp rate lets it reach. A unit that the plan has committed starts at its minimum stable load. Next comes the output of units that must run. Then come wind and solar, at the output that the weather gives. A stack of offers, sorted cheapest first, covers the rest of the load. Storage bids into that stack at the value of the energy that it holds. A unit on Reserve goes last in the stack. Its bid does not change.

A plant bids the cost of its next MWh. That cost has three parts. The first is the fuel price divided by the efficiency of the plant. The second is its carbon intensity times the carbon price, also divided by the efficiency. The third is the variable operating cost. Efficiencies come from the catalogue. Research raises them, up to a cap of 95%. The catalogue lists a combined cycle plant at 59%, a supercritical coal plant at 45%, an open cycle gas turbine at 36% and a reactor at 34%. On the baseline world, the first fuel prices are $22 per MWh thermal for gas, $9 for coal and $5 for uranium. Municipal refuse is at minus $4: the plant gets a gate fee to take it. Gas releases 0.202 tonnes of carbon dioxide per MWh thermal. Coal releases 0.34 tonnes. The internal carbon price on the Operating policy screen goes up to $300 a tonne. It changes only the merit order. You never pay it.

The minimum load makes the merit order more than a sorted list. A combined cycle plant cannot run below 30% of its rating. A reactor cannot run below 55%, and a refuse plant cannot run below 60%. The rating is what the unit can do in its current condition. It is not the nameplate output. So the dispatcher reads the stack in two passes. The first pass takes only units whose minimum load fits in the load that is not yet covered. The second pass starts the units that the first pass held back, smallest minimum first. Without this pass, customers lose power. This pass can start only a unit that can start within the 15 minutes. A slower unit that is off offers nothing until the plan starts it. If the colony still has a shortfall, the dispatcher orders every unit that is off to start at once. Each unit comes online after its full start time. A refuse plant must run, because the rubbish keeps arriving. You can also set any unit to Must run by hand.

In a surplus, storage fills first. Then air capture on Spare power takes what is left. After that, the electric boilers of the towns take power. Then the dispatcher reduces the output of plants, highest bid first. It reduces thermal plants first. Their must run floor and their ramp rate limit how far it can reduce them. For a slow unit that the plan needs again soon, the minimum stable load is also a limit. Because wind and solar bid almost nothing, they are among the last to be curtailed. A dam at its spill line ranks with wind and solar. A plant that burns fuel that arrives anyway also ranks with them. When the dispatcher handles a surplus in this way, the price is $0 per MWh. If the fleet still makes power that it cannot stop, that power is spilled into dump banks. You pay $30 per MWh to dispose of it. The price for that step is minus $30 per MWh. In a shortfall, the heat pumps of the towns use less power. Then the stores release their emergency reserves: first the batteries, then the dams and the solar towers, then the hydrogen of turbines that cannot buy fuel at other times. A hydrogen turbine buys fuel if the operating policy lets it. The rest is unserved energy. There is no interconnector. No power comes in from outside, and you cannot sell a surplus.

  • Ramp rates are modelled in both directions. Each plant has a ramp rate as a percentage of its rating per minute. This rate sets how far the plant can move up or down in one 15 minute step. A reactor at 0.5% a minute moves 4.5 MW of its 60 MW. An open cycle turbine at 14% can cross its whole range in one step. The card of a plant shows the ramp rate only where it limits the plant.
  • The turbine of a dam can cross its range in 3 minutes. Its ramp rate never limits it. Two other things limit it. The first is a minimum load of 40%: below this load, the turbine vibrates. The second is a minimum flow that its licence makes it release at any price: 15% of the river. This water goes through the turbine, so a dam makes a little power also when wind or solar power is curtailed. If the turbine stops, the same water goes over the spillway. Lines connect to a dam at its powerhouse only, never on the barrage.
  • Unit commitment models start times, start costs, and minimum up and down times. A plan for the next 48 hours selects units by their dispatch value. It commits 15% more than the forecast load. A slow unit stays on for at least its start time, and never for less than 1 hour. A start soon after a shutdown is a hot start. It costs 0.4 of a cold start. Ramp rates, minimum load and physical availability still limit the actual output.
  • The system marginal value is an internal dispatch signal. It is not a payment between your own plants. It is the highest bid that the dispatcher takes. So a refuse plant that gets a gate fee can push it below zero. Customers on a fixed price pay the tariff. Hourly tariffs pay the system signal plus the stated margin, between $0 and $1,000 per MWh. The revenue shown for each asset is only an allocation.
  • You can build a colony as separate grids with no line between them. Then the dispatcher plans each grid on its own units, against its own load and losses. Each grid has its own system marginal value. The published price is the price of the largest grid. When protection opens a circuit, the circuit splits a grid in the same way until it closes again.

The colony is one integrated utility. All generation and storage have one goal: to meet demand at the lowest expected operating cost. They stay within the operating policy and within what the units can do. Thermal costs include fuel, efficiency, carbon and variable maintenance. The same rolling scheduler gives a value to stored water, battery energy and hydrogen. You do not pay this value as a cash cost. A store pays for its power once, when it charges. Conversion losses and cycle wear affect dispatch. Fixed costs and capital recovery go into the cost of supply and the tariff. They do not go into the merit order.

The forecast covers 14 days. It is hourly on the first day, in blocks of 3 hours on the second day and in blocks of 6 hours after that. It uses the connected generation, the current fuel costs, the daily load shape, and weather that slowly returns to the seasonal normal. Each store has its own limits for energy, power and conversion. Two scheduling passes share the rest of the load across the stores. Identical batteries share one forecast. In dispatch, each battery keeps its own limits. This is an approximation. It is not a fully optimal unit commitment. Because of seasonal targets, long duration stores do not empty too early. Plans refresh every 6 hours. They also refresh when the weather, the river or the load moves away from what they assumed, or when the fleet changes. When you save and then load the game, the plans stay. Plans never use past electricity prices.

Inside the scheduler, unserved energy has a value of $10,000 per MWh. This planning value is not a market price. The utility never pays it. The published price never goes above $3,000 per MWh. A step with a supply shortfall publishes exactly that cap. When a network limit cuts power to customers, the price does not change. Hourly tariffs pass the published price to customers between $0 and $1,000 per MWh. Dispatch enforces network limits and the minimum output of thermal plants. The forecast does not optimise them. The forecast also simplifies future heat demand, the behaviour of hydrogen units at part load, and the production of synthetic methane.

Demand and money

The game builds the load tile by tile. Every developed tile behind a Local Substation holds dwelling equivalents. Each one stands for 3.23 real dwellings. Before space heat and car charging, each one draws an average of 1.15 kW on residential, 2.35 kW on commercial and 4.4 kW on industrial ground. The game mixes nine shapes of end use across twelve kinds of area. So a commuter suburb, a business district and a factory with two shifts each have their own daily curve. A street of houses that empties on weekdays has its peak at hour 19, at 1.95 times its daily mean. A process that runs 24 hours a day has its peak at 1.07 times its daily mean.

Homes and shops start to heat when the outside temperature is below 15.5 °C. Industry starts below 13 °C. Cooling starts above 22 °C. So the seasonal shape follows the weather of the colony. On the standard world, demand grows 2.2% a year. Once a month, the game updates the population of every town. The update uses how well you supplied the town, public opinion, the price of power and energy poverty. The first town starts with 9,000 to 11,500 people. Another town is founded 3 to 5 years into the run. After that, a new town comes every 3 to 6 years. A new town appears as an outline on the map, with a marked site for its 110/20 kV substation. Nothing is built on it. When you build the substation on the site, you become responsible for the town. From then, you have 12 months to connect its streets to Local Substations. After that, the town fines you for each street that still waits.

You set one flat tariff. It starts at $118/MWh. You can also offer an hourly tariff: the published price plus a margin. The regulator counts the cost of supply plus $22/MWh as the fair price. It judges your tariff by how far it is above the fair price. Customers react to a slow average of past prices. They do not react to the current price. They use less power with an elasticity of minus 0.34. The reference is 0.8 of what local incomes can afford: $120/MWh in the first year. Cheap power brings people to a town. Above about 0.95 of what incomes can afford, people fall into hardship. They stop paying. At full hardship, customers never pay 42% of their bills. If you keep your tariff more than $42 above the fair price, the regulator caps it for 5 years. It also fines you $15 million or 0.6 of the revenue of that year, whichever is larger.

Revenue is the delivered energy times the tariff times the collection rate. The costs are fuel, carbon, start costs, variable and fixed O&M, grid upkeep, disposal of unsold power at $30/MWh, and interest. Credit has one fixed limit. On the standard world, it is $120 million at 4.5%. You start with $180 million of cash. The limit never changes. There is no credit rating. Nothing is repaid automatically. If your cash is used up and your credit line is fully drawn, the run ends in administration.

You pay the carbon price on every tonne that the energy use of the colony emits. This includes power plants, boilers in homes, shops and works, the local heat supply of the heat networks, and petrol and diesel vehicles. The game counts industrial processes and farming, but you do not pay for them. Captured tonnes are subtracted from the bill. If the colony captures more than it emits, you get the carbon price for the difference.

  • Fuel prices at the plant gate when the colony is founded, in dollars per MWh thermal: coal $9, gas $22, oil $45, uranium $5, biomass $15, hydrogen $120, and waste at minus $4. The minus price is a fee that the plant gets to take the waste away.
  • After that, each fuel price drifts. It has random swings. From time to time, it has a shock that lasts several months. Gas drifts up 1.7% a year at a volatility of 0.28. Hydrogen drifts down 2.8% a year.
  • On the standard world, the carbon price starts at zero. Then it rises: $4 a tonne at year 8, $22 at year 15, $65 at year 25, $130 at year 40 and $190 at year 60.
  • Each plant has its own build time: 6 months for a wind turbine or a solar park, 24 months for a combined cycle gas plant and 60 months for a small modular reactor.
  • Households pay retail prices for their own equipment. Rooftop solar costs $2,000 per kWp. A heat pump costs $10,500 more than a new gas boiler. An electric car costs $14,000 more than the petrol car that it replaces.
  • Road fuel has a flat duty of $42/MWh thermal, in addition to the wholesale crude price and the supplier markup. This duty is a large part of the reason that an electric car earns back its cost.

Weather, wind and solar

The game makes all of its weather. Temperature, cloud and wind each follow the season along one cosine over the year. Each also has an anomaly: a departure from normal that fades with a fixed persistence. The persistence is 46 hours for wind, 30 hours for cloud and 70 hours for temperature. Because of this, weather lasts for days and does not change every step. Heavy cloud with weak wind counts as a blocking high. A blocking high cuts the wind target to 32%. In winter, it also lowers the temperature by up to one standard deviation of its anomaly. This is about 3.7 K to 5.2 K. The value changes with latitude. So a drop in wind and solar and a peak in heat demand come from the same weather system.

One curve changes wind into power. Below 3 m/s, a turbine makes nothing. Above that, the output follows the cube of the wind speed at hub height, up to the full rating minus 1% for transformer and cable losses. Above 25 m/s, storm control ramps the output down to zero at 30 m/s. The turbine does not stop at once. The rotor sets the wind speed at which a turbine reaches its rating. This speed is the cube root of the rating divided by the swept area, half the air density and a power coefficient of 0.45. The 6 MW turbine with a 162 m rotor reaches its rating at 10.18 m/s. The 15 MW offshore turbine with a 236 m rotor reaches its rating at 10.76 m/s.

Wind also changes across the map. On new maps, the ground around a site sets its wind quality. Open water adds the most. The first row of land on a coast gets a smaller increase. This increase fades over about 3 tiles inland. Woods reduce the wind quality a little. Rivers and ponds add nothing. Each plant then runs on the wind at its own position. On average, this is 0.865 of the wind speed of the climate.

On colonies founded after weather fronts came into the game, weather also crosses the map at 35 km/h. Each site gets the wind and cloud that the upwind edge of the map had earlier. At the far corner, this is up to about 2.5 hours earlier. The direction turns with the weather. The climate stays the same. Only the timing changes. So wind farms at opposite ends of the map rise and fall at different times. Together, two farms 57 km apart vary nearly a fifth less from hour to hour than one farm alone. Without fronts, the difference was less than a tenth.

Solar starts from the position of the sun. Declination and hour angle give the elevation of the sun. Clear sky irradiance is 1120 W/m2 times the sine of that elevation, reduced for air mass. At mid latitude, it has its peak of 776 W/m2 at noon on day 172. With full overcast, 22% of that still reaches the panels. Output is the irradiance divided by 1000, times a ground factor from 0.55 to 1.30, times a loss of 0.4% per kelvin above 25 °C, times 0.96 for the rest of the system. At full cover, snow removes up to 70% of the output of a plant.

Storms show on the weather panel in the top bar. There are no alerts for them. At mid latitude, 4 to 5 storms come each year. Each storm lasts 5 to 22 hours, with the wind at 16 m/s or more. At full strength, the wind reaches 32 m/s. That is above 30 m/s, the speed at which the turbines stop. A harsh world multiplies the hourly chance of a storm by 1.4. The game once showed alert cards for storms, dark doldrums, cold waves and heat waves. They were removed for two reasons. Weather that lasts for days belongs on a display. Also, at 400× speed, a card came back every few seconds.

  • A capacity factor is the share of its nameplate output that a unit averages over a year. A turbine at 36% delivers 36% of its full output over all the hours of a year.
  • For a normal world, the game draws the latitude between 0.33 and 0.6. For a harsh world, it draws it between 0.62 and 0.78. The figures here are for 0.45. At this latitude, the mean wind of the climate is 7.28 m/s.
  • At that latitude, the 6 MW onshore turbine averages 25.1% of nameplate output on the poorest ground that it accepts, which is site quality 0.35. It averages 36.5% on middling ground and 52.1% on the best ground.
  • A solar park averages 10.4% of nameplate output on the poorest ground that it accepts, which is site quality 0.3. It averages 17.4% on the best ground.
  • The wind planning figures come from the power curve over a Rayleigh wind distribution in 48 bins. The solar figures come from the path of the sun over 24 sample days of 24 hours each. Three simulated years of live weather on three seeds give 36% to 38.5% for wind and about 13% for solar at a middling site. The planning estimates are 36.5% and 13.4%. So the two agree within two points.
  • Site quality goes from 0.62 to 1.34. It multiplies the wind speed. It does not multiply the output. So better ground is worth more than its number suggests.
  • Each year has a wind index, a solar index and a river index. The seed and the year fix them. Wind is between 0.75 and 1.25 of normal. Solar is between 0.82 and 1.18 of normal. The planning briefing shows an estimate of the index for that year. When you research better forecasts, the estimate gets more accurate.

Voltages, lines and congestion

There are four voltage levels: 380 kV, 110 kV, 20 kV and 400 V. Power moves between them only through a transformer. A rated transformer is modelled as two busbars with a winding between them. So power that goes down to 20 kV and back up again loads both transformers. The ratings are for a colony with a peak of about 60 MW. That is why the numbers below are small.

The rating of a circuit is its catalogue rating, raised by 8% and then by another 5% through conductor research, and scaled by its condition. A damaged line carries 0.35 plus 0.65 times its health. So at zero health, it still carries about a third. Losses grow with the square of the flow. One map tile is 600 metres of route. Susceptance (kilovolts squared over ohms per kilometre) sets how the flow splits between parallel paths. The voltage and the conductor set the susceptance. The rating does not. So a 20 kV feeder beside a 110 kV circuit has about 1/27 of the susceptance of that circuit. It carries almost none of the through flow. The game solves each electrical island as a DC load flow. When the network changes, the game factorises it with a sparse Cholesky method. A conjugate gradient solver is the fallback.

Congestion is a line or transformer that carries more than its rating. First, the batteries that charge behind it charge less, in up to six rounds. After each round, the game solves the flow again. The size of each round depends on how much the last round helped. Plants outside the congested area reduce their output by the same amount. If a round does not help, the game reverses it. Next, the heat plant of the towns behind it uses less power. Electric boilers that run on surplus go first. Then come heat pumps that fill heat stores, and then the other heat pumps. If the line or transformer is still above its rating, the game cuts power to customers downstream. To find them, it follows the power from the overloaded line or transformer in the direction of the flow. It never goes back through the overloaded line or transformer itself. The game does not cut everything that is connected. In a meshed grid, that cuts the whole colony, because the grid is one island. The game books each cut to the network that caused it: distribution, transmission or the 380 kV backbone.

You can run a line above its rating on purpose. For one circuit, use its card. For every circuit at once, use the Control room. The line can then carry up to 1.28 times its rating. While it is above its rating, it loses 0.006 condition an hour. It gets no repairs until it is back below its rating. An overloaded line gets hotter with the square of its loading. At 41% above its rating, it reaches its limit in about 30 minutes. Protection then opens it for 45 minutes. This costs 0.04 condition. Protection closes the line again while it is still half warm.

You can switch any line to Inactive and back to Active by hand. A rule can also switch it automatically. The rule acts when the loading of a watched circuit passes a set level, or for a range of months. A loading rule switches the line back only after a flow check. The check must show that the watched circuit stays 10 points below that level.

Storms damage overhead lines. In a strong storm, each line has a chance of damage that grows with its length. A hit removes 0.3 to 0.85 from its condition. Crews repair it at 0.007 of condition an hour. So a damaged line is back in service within about a week. Transformers lose a fixed amount of power as soon as they are energised. They also lose a share that grows with the square of their loading. The 110/20 kV substation of a town starts at 25 MW. Each transformer bank adds 25 MW, up to 175 MW. A 380/110 kV substation takes up to three extra banks of 25% each. You can reinforce a 20/0.4 kV Local Substation four times, by 50% of its original rating each time.

A tile along a route is a junction when another route of the same voltage is also on that tile. The exception is a route that runs next to it on both neighbouring tiles. So two circuits that share a corridor connect where they meet and where they separate. They do not connect at every tile between those points. The flow solver, the split of a tapped route and the map all use this one rule. So the drawn line and the simulated circuit always match. The ends of a route are never junctions. An end on a tile with nothing else on it shows as loose.

  • 380 kV overhead line: 1800 MW at 0.042% loss per kilometre and 0.26 ohm per kilometre. The model uses a 265/35 circuit with bundles of four conductors.
  • 110 kV line: 80 MW for a single circuit, 160 MW for a double circuit and 320 MW for a quad circuit, at 0.19%, 0.16% and 0.14% loss per kilometre.
  • 20 kV feeder: 14 MW for one cable, 28 MW for two cables and 56 MW for four cables, at 1.04%, 0.88% and 0.80% loss per kilometre.
  • The 20 kV feeder and the lightest 110 kV circuit use the same AL/ST 120/20 conductor: 14 MW at 20 kV and 80 MW at 110 kV. The same wire carries almost six times as much at the higher voltage.
  • Transformers: the first town starts with a 110/20 kV substation of 25 MW. This is the same substation that you build for every later town. Each bank adds 25 MW, up to 175 MW. The 380/110 kV substation passes 300 MW, and the grid node passes 600 MW.
  • A 20/0.4 kV Local Substation has no throughput limit in the network model. Its rating is 6 MW for its own streets, within 3 tiles.

Storage and hydrogen

Each kind of storage holds energy for a different length of time. The Grid Battery is 40 MW and 120 MWh, which is 3 hours. It gives back 88% of what it takes. The Long-Duration Battery is 25 MW and 250 MWh, which is 10 hours. It gives back 60%, because a flow battery loses energy to its pumps and to internal currents. Hydrogen can hold energy for a season, but it gives back the least. Every store loses a little energy each day: 0.8% a day for a full lithium battery, down to 0.006% for a salt cavern. Both batteries and the electrolyser wear with cycles and age, down to a floor of a third of their nameplate rating.

A surplus goes to the stores in a charging order that you set. What they cannot take is curtailed. Outside a surplus, stores charge only when the net load is a set margin below its average over 2 days. They charge only up to that average. So the stores fill dips and do not make new peaks. They also charge only from plants whose bid is below the value of the stored energy after losses. A store discharges at a bid. The bid is the value of its energy from the 14 day plan, times a policy setting with a default of 1.0, plus variable O&M, plus wear. Wear is the capital cost divided by the warranted throughput. It is $18.33 per MWh for the lithium battery and $2.27 for the flow battery. By default, each battery keeps 15% as an emergency reserve.

Every store whose energy can stop a shortfall can keep an emergency reserve: a battery, a hydrogen store, the lake of a dam and the salt tank of a solar tower. The card of each store sets its own share. A store without a setting follows the operating policy, which gives 15% to batteries and none to hydrogen, dams and solar towers. Energy below the line is not sold, burned or planned. The dispatcher uses it only in a shortfall, or when you release the reserves in the control room. A dam below its line runs only its river.

You can give each battery a role. Fleet follows the operating policy. Backup stays full for a shortfall. It never sells into the market, and it recharges at any price. Spare power charges only from power that is otherwise curtailed or spilled.

The Electrolyser Plant draws 60 MW and changes it into hydrogen. Its efficiency is 72% at full output. Its best efficiency is 76.9%, at 30% of full output. It keeps a quarter of its 600 MWh buffer on site. It sends the rest through pipes to the storage vessels, smallest first. So a 900 MWh tank farm fills before a 40,000 MWh cavern. The Hydrogen Turbine burns the hydrogen at 40% efficiency. So power to hydrogen and back to power returns 28.8%. The same 14 day plan as for the batteries gives a value to stored hydrogen. The policy for the turbine has three settings. With Own hydrogen only, it burns only hydrogen that the colony made. With Only when short, it buys on the market only when the colony otherwise has a shortfall. With Whenever it pays, it buys whenever bought hydrogen is the cheapest way to cover the hour. Bought hydrogen costs the market price: $120 per MWh thermal in the first year.

Storage can stay idle while wind is curtailed. This was reported as a bug. A test over 120 simulated days in 15 minute steps found that it is correct. There are two causes. The tests of the game treat anything else as a fault.

  • An electrolyser has a minimum load of 15% of 60 MW. So it needs at least 9 MW to run. Below that load, hydrogen crosses the diaphragm, and the gas mix becomes unsafe. If the surplus is less than 9 MW, the electrolyser stays off, and the surplus is curtailed. The colony does not start a gas plant to keep the electrolyser on.
  • When hydrogen storage is full, the electrolyser stops. It keeps a quarter of its buffer on site and moves the rest into the vessels. When the tanks and caverns are full, its buffer also fills. Then there is no place for more hydrogen to go.
  • The tank farm and the cavern are not on the electrical grid. A pipe from the electrolyser fills them. They supply no power. So a line to one of them does nothing.

Heat networks, hydrogen and air capture

Towns create and run their own heat networks. You do not build the pipes. The heat operator of a town orders large heat pumps and electric boilers through contracts. When you sign the contract, the plant is built. You connect it to the grid. For that power, the operator pays the published price of the hour, between $0 and $1,000 per MWh, plus $22. It always pays in full. A heat pump runs only in the hours when its heat is cheaper than the local heat supply of the network. Electric boilers take surplus power after air capture.

Hydrogen that the colony does not need goes into the gas grids of the towns as a blend. A town can also offer a supply contract. The contract is for a fixed amount of hydrogen a year, for 5 years, or for fewer years late in a run. The amount is about 40% of what the gas grid of the town can blend. If you deliver less by the end of a year, the town buys the missing hydrogen at the market price. It bills you that price, plus a fine of $40 for each missing MWh. A Methanation Plant changes 1.25 MWh of hydrogen into 1 MWh of methane. The existing gas plants burn this methane with no net carbon.

Direct air capture runs in one of three modes. Always runs 24 hours a day. Cheap hours runs in the cheapest third of the day, by the forecast price. Spare power runs only on power that is otherwise curtailed or spilled, after the stores are full.

Plants, condition and end of life

Everything that you can build is one catalogue entry. The entry has these physical figures:

  • capacity in megawatts
  • efficiency: the electricity out divided by the fuel heat in
  • capital cost for the whole unit, not per kilowatt
  • fixed operations and maintenance per year
  • variable operations and maintenance per MWh
  • months to build
  • design life in years

The list price falls along a learning curve as more of that technology is built. It also falls more slowly as the years pass. It does not fall below a floor of 25% of the list price. You pay fixed maintenance for every hour that the asset exists, also in the hours when it generates nothing. A mothballed unit still pays 40% of it.

Condition is one number that multiplies the output of a unit. It covers both wear and damage. Age sets a ceiling. Condition can never rise above this ceiling. Without maintenance, the ceiling falls by 0.45 at an even rate over the design life. So it is at 55% when the design life ends. It stops falling at 0.25. Damage is the gap below the ceiling. Crews can repair damage. Readouts compare condition with the ceiling for the age of the unit. They do not compare it with a new unit. An overhaul restores condition to a ceiling that also falls with age. This ceiling falls by 0.28 across the design life, and by another 0.45 for each design life after that. Its floor is 0.35.

Three things set the failure risk: a base yearly rate for the kind of unit, the running hours since its last service, and the condition that it already lost. In good order, the base rate is 1.37% a year for a fuelled plant, 1.9% for a wind farm and 0.5% for solar. One service interval is 30,000 running hours. A full interval multiplies the base rate by 3.9. A fuelled plant that trips falls to between 25% and 60% of its ceiling. Any other unit falls only to between 66% and 88%. The reason is that one broken gearbox in a farm of twenty turbines only reduces the output of the farm. Repairs restore 0.013 condition an hour for a fuelled plant, 0.009 for most other assets and 0.005 for a wind site. You get a card only when the loss is at least 1 MW and more than the firm headroom that is left to cover it. After you unlock the Control room, you also get a card when a trip makes underfrequency protection cut power to customers. You book overhauls into a month on the planning screen. A crew that you send outside that plan costs 1.45 times as much.

A demolition gives 3 months’ notice. At once, the asset no longer generates or carries power. The crews remove it 90 days later. You get scrap worth 12% of its current build cost only at removal. So if you cancel within the notice period, it costs nothing. A retirement that you book on the planning screen removes the unit in the chosen month instead. A rebuild in place costs 65% of a new build and takes the full build time. Until the crews finish, the old unit stays in service in its current condition. The rebuild resets condition, ceiling, running hours and age. It keeps the name of the asset, every line connected to it, its town and the programmes bought on the streets of that town. You can also repower a wind, solar, hydro or geothermal site after 70% of its design life. A repower costs 55% of a new build and takes 45% of the build time, with a minimum of 3 months.

  • The Wood-Fired Steam Plant is 18 MW at 22% efficiency. It costs $26,000,000 and takes 8 months to build. Its design life is 30 years.
  • The Combined-Cycle Gas Plant is 100 MW at 59% efficiency. It costs $75,000,000 and takes 24 months to build. Its design life is 35 years.
  • The Small Modular Reactor is 60 MW at 34% efficiency. It costs $300,000,000 and takes 60 months to build. Its design life is 60 years.
  • The Wind Turbine is 6 MW, with a rotor of 162 metres. It costs $9,400,000 and takes 6 months to build. Its design life is 25 years.
  • The Hydroelectric Dam is 25 MW and uses no fuel. Its reservoir holds a week at full output. It costs $72,000,000 and takes 30 months to build. Its design life is 80 years.

The power plants page shows every unit that you can build, with these figures.

What it does not model

These are the limits of the model. Some of these things were never built. Others were built, tried and removed, because they took more attention than they were worth. When a mechanic was removed, its figures stayed in the catalogue. So it can return without a new measurement of every unit.

  • Unit commitment is modelled. A unit that stops needs its full start time before it is online again. Each start costs money and service life. A unit must stay off for a minimum time before it can start again. After it starts, it must stay on for a minimum time. Some units can start within the 15 minute step, such as a gas turbine or a dam. Such a unit responds within that step and still pays for the start. That fast response is the main reason to own one.
  • So the dispatcher plans ahead, like a real control room, because a slow unit must start before it is needed. It runs the merit order hour by hour over the next 2 days. The load for this plan is what the fleet must cover: the daily load of the colony, minus the expected wind and sun, minus what the stores plan to deliver or draw. A slow unit starts in time to be online for the first hour that needs it. Between those hours, it stays at its minimum if the plan needs it again before it can restart. It also stays at its minimum if that costs less than a restart. Otherwise, it stops. The merit order switches a unit that starts within one step on and off every step. After you unlock the Control room, the plan also keeps a spare firm unit to cover the loss of the largest one. A battery can respond within 1 second and counts as that reserve.
  • There is no interconnector and no neighbour to trade with. A shortfall becomes unserved energy. A surplus is spilled into dump banks, and you pay for it. The game books each shortfall with its cause: too little plant, a plant at its ramp limit, a plant in its start time, or the network between the plant and the load.
  • The network carries only real power. It is a DC load flow. So it has no reactive power, no power factor, no voltage magnitude and no tap changer.
  • Every synchronous generator in service adds spinning mass. The game shows the total as seconds against the load. Wind, solar and batteries add none, however much they generate, because they connect through converters. From that total, the Control room shows how fast the frequency falls if the largest unit in service trips now. After you unlock the Control room, such a trip is a real event in the game. The game calculates the loss for the step. If the fast response cannot stop the fall, underfrequency protection cuts power to customers in stages. If protection must cut more than about half the load, the result is a cascade: a blackout of the whole colony. A unit that is off cannot stop the fall. But it ends the blackout sooner. As each unit starts, more towns get power again. With a gas turbine or a dam, the towns have power again within about 20 minutes. Without one, the restoration takes about 35 minutes. The load that protection cuts counts as blackout minutes. It does not change the unserved energy. Before you unlock the Control room, a trip only reduces output. The frequency reading moves only in a shortfall.
  • The Control room shows the operating reserve. The plan includes it automatically. There is no setting for it. Response is the power that can react within 30 seconds: governor droop on synchronised generators, and inverters. Headroom is what the fleet can add within one step, after ramp limits. Units that are off do not count. The game reads both from the fleet. After you unlock the Control room, the plan keeps enough firm plant in service to survive the loss of the largest unit. It skips a unit on Reserve. It also skips a unit whose minimum load pushes the colony into a surplus. The Control room then shows N-1 short. There is no reserve setting, because a reserve that the dispatcher ignores is worse than no reserve.
  • There are no wake losses between turbines. So ten turbines in a row make ten times what one turbine makes.
  • Solar uses only horizontal irradiance. One ground multiplier includes panel tilt, direction and tracking.
  • Hydrogen is a single stock for the whole colony, measured in MWh thermal. There are no kilograms, no pressures, no compressors as separate units and no pipe network.
  • Every route is an overhead line. Buried cable was in the catalogue and was removed. Pumped storage was also removed, because it did nothing that the battery did not already do.
  • A breakdown is a fall in one condition number. The game does not model the failure of a named component. There is no parts inventory, no spares lead time and no crew roster.
  • There is no corporation tax and no general inflation. Every price is in the same dollars for 60 years. The exceptions are the fuel drifts, the carbon path, the technology learning curves and incomes, which rise 1.6% a year.
  • Capital costs are approximate. The efficiencies, carbon intensities and minimum loads that a bid uses are real current figures. The prices of wind, solar, batteries and hydro are close to what they cost today. Gas plants are at the cheap end of real prices. Coal, wood and nuclear plants cost much less in the game than in Europe or the US.