Ask an engineer from the 1990s how to plan a power system and you would hear about three layers. Baseload plants ran day and night. Mid-merit plants followed the daily rise and fall. Peakers came on for the hottest afternoons or coldest evenings. That model fitted a world of coal, nuclear and gas, where the only thing that varied was demand. Wind and solar have added a second source of variability on the supply side, and the old layers are being reshuffled. This guide explains the classic terms, the technical limits of each plant type, and why “flexibility” has replaced “baseload” as the word planners care about most.
Reading a load curve
Demand on a grid follows a strong daily pattern. In a temperate country it is lowest around 3–5 am, rises through the morning, plateaus during working hours and peaks in the early evening when lighting, cooking and heating overlap. In hot countries like Malaysia, air conditioning pushes the peak into the afternoon. Weekends are lower, and there are seasonal swings driven by temperature.
Sort every hour of a year from highest to lowest demand and you get a load duration curve. Its shape tells you how much capacity is needed for how many hours:
| Band | Hours per year needed | Classic plant | Why |
|---|---|---|---|
| Baseload | 6,000–8,760 | Nuclear, coal, lignite, run-of-river hydro | High capital cost, low fuel cost, best run constantly |
| Mid-merit (intermediate) | 2,000–6,000 | Combined-cycle gas, newer coal | Moderate costs, can follow daily swings |
| Peaking | Under 1,000–2,000 | Open-cycle gas, gas engines, oil, reservoir hydro | Cheap to build, costly to run, fast start |
The logic is economic. A plant with high fixed costs and low fuel costs needs to run many hours to be cheap per kWh. A plant with low fixed costs can afford to sit idle most of the year. The capacity factor of each band follows directly; test it with the capacity factor calculator.
How flexible are different plants?
Every thermal plant has physical limits: how fast it can change output (ramp rate), how low it can go before it must shut down (minimum stable load) and how long it takes to start. Typical values from IEA and Agora Energiewende studies:
| Technology | Ramp rate (% of capacity per min) | Minimum stable load | Hot start time |
|---|---|---|---|
| Nuclear (load following, e.g. French PWRs) | 2–5% within 50–100% band | ~50% | Many hours to days from cold |
| Lignite | 1–2% | 40–60% (older), ~35% (retrofitted) | 2–6 h |
| Hard coal | 1.5–4% | 25–40% | 2–4 h |
| Combined-cycle gas (CCGT) | 2–8% | 30–50% | 30–90 min |
| Open-cycle gas turbine (OCGT) | 8–15% | 20–50% | 5–15 min |
| Gas engines | 15–20%+ | ~10–30% per engine | Under 5 min |
| Reservoir hydro | Up to full output within minutes | Low | Minutes |
| Battery storage | 0 to 100% in under 1 s | None | Instant |
Ranges vary with plant age and design. Older coal plants are much stiffer than retrofitted ones, and Danish and German coal plants were modified in the 2010s to run down to 20–25% load as wind grew. France routinely varies nuclear output: its reactors were designed to follow load because nuclear supplies around two-thirds of French electricity. See How a Nuclear Reactor Works for the physics behind that.
Net load and the duck curve
With wind and solar on the grid, what thermal plants actually have to supply is net load: demand minus variable renewable output. On a sunny spring day in California, net load falls sharply after sunrise, bottoms out around midday and then climbs steeply as the sun sets and people come home.
The California Independent System Operator (CAISO) first published its “duck curve” chart in 2013 as a forecast. Reality overshot it. Midday net load has fallen far below the original projections on spring days, and the evening ramp, the climb from the midday low to the evening peak, has exceeded 20 GW within about three hours on some days. Batteries changed the picture from 2020 onwards: California went from almost no grid storage to well over 10 GW of batteries by 2024, much of it charging at noon and discharging into the evening ramp.
At 3% per minute, a 500 MW CCGT adds 15 MW/min, so the ramp alone needs about 6 such units ramping flat out, or batteries.
Why “baseload” is losing its meaning
When wind and solar have near-zero running cost, they are dispatched first. In sunny or windy hours, there may be no room left for a plant that wants to run at constant output. The grid then has a choice: curtail renewables, throttle the inflexible plant, export, store or find new demand. In South Australia, rooftop and utility solar have at times supplied more than 100% of the state’s demand on spring middays, and minimum operational demand has fallen below zero on some days.
None of this means firm capacity has stopped mattering. A grid still needs enough dependable supply for a cold, still evening or a cloudy week. What has changed is the job description. The valuable resources are the ones that can:
- ramp quickly from low to high output,
- start and stop often without damage,
- run economically at low capacity factors,
- shift energy in time (storage),
- or change demand (see Demand Response Explained).
Planners now talk about firm, flexible and variable resources instead of baseload, mid-merit and peak.
Flexibility toolbox
| Source of flexibility | Timescale | Notes |
|---|---|---|
| Batteries | Seconds to ~4 hours | Fastest, falling costs, limited duration |
| Pumped hydro | Hours to days | Large, site-limited; see Hydropower Explained |
| Reservoir hydro | Hours to seasons | Excellent where available |
| Gas turbines and engines | Minutes to weeks | Flexible but fossil and exposed to gas prices |
| Interconnectors | Minutes to seasons | Share weather across regions |
| Demand response | Minutes to hours | Cheap if programmes are well designed |
| Retrofitted thermal plants | Hours | Lower minimum loads, faster ramps |
The picture in Malaysia
Peninsular Malaysia’s grid has been dominated by coal and gas, with coal units supplying much of the steady base and gas plants following load. Peak demand has climbed with economic growth and air conditioning use, reaching about 19–20 GW in recent years according to Grid System Operator data, with the peak typically on hot weekday afternoons. As large-scale solar and rooftop solar expand under the national roadmap, the afternoon peak will be partly met by solar and the stress will move to the early evening, the same shift seen in California and Australia. Plans for battery storage and flexible gas capacity reflect that. Read more in How Malaysia’s Electricity System Works and on the Malaysia electricity page.
In practice
- Baseload, mid-merit and peaking describe how many hours a plant is economic to run, not a fixed property of the technology.
- Ramp rates and minimum loads limit how well thermal plants work alongside wind and solar.
- Net load, not demand, is what dispatchable plants must follow, and its evening ramp is the new design challenge.
- Flexibility from storage, hydro, interconnectors and demand is now worth more than constant output.
To compare what each technology costs per MWh over its life, try the LCOE calculator and read LCOE Explained.
Calculators for this topic
Frequently asked questions
What is baseload power?
What is a peaking power plant?
Do we still need baseload plants with lots of solar and wind?
What is the duck curve?
How fast can different power plants ramp?
What is minimum stable load?
Sources & references
- IEA (2018). Status of Power System Transformation: Power plant flexibility
- Agora Energiewende (2017). Flexibility in thermal power plants
- CAISO: What the duck curve tells us about managing a green grid / Today's Outlook data
- OECD-NEA (2011). Technical and Economic Aspects of Load Following with Nuclear Power Plants
- Grid System Operator Malaysia: system data
Figures in this guide are typical ranges drawn from the sources above; your own numbers depend on location, prices and equipment. Use the linked calculators for site-specific estimates.