Honglin Li
  • Research
  • Projects
  • Publications
  • Software
  • Teaching
  • Learn
  • Notes
  • CV
  • Journeys

Why does electricity demand rise and fall?

Read a real year of demand, then add solar and dig the duck curve

Lesson 2 of Energy Systems, Step by Step: read a real year of hourly electricity demand, scaled to a small fictional town: a hot day against a cold one, weekdays against weekends, the seasons, a load-duration curve built by sorting 8,760 hours, and the duck curve that solar digs.
  1. Learn
  2. Lesson 2
  • Foundations · L1
  • Lesson 2 of 14
  • 15 min
  • After lesson 1
  • Completed
  1. 1Step by step
  2. 2Explain
  3. 3Check
  4. 4Sources

Two real days of 2025, scaled to Prairie Town (MW)

Prairie Town's demand on two real days of 2025. 29 July 2025: one hump, from 298 MW at 04:00 to 450 MW at 16:00. 21 January 2025: two humps, 389 MW at 07:00 and 359 MW at 19:00. 010020030040050000:0006:0012:0018:0024:00
29 July 202521 January 2025

With JavaScript on, you compare these days, browse every week of 2025, sort the year's 8,760 hours into a load-duration curve, and add solar to dig the duck curve.

Data. Processed hourly profiles of the Southwest Power Pool (SPP) in 2025 from EIA-930, scaled to a fictional town whose busiest hour is 450 MW, as in lesson 0.
Repairs. Some demand and solar hours were rebuilt; see “What the data leave out” below.
Hours. Central time; 16:00 means 16:00 to 17:00.

What you just did, as diagrams

Lesson 0 ran one day like these hour by hour; lesson 1 showed that the area under a power curve is energy. Here the curve itself is the subject.

Daily shape
On 29 July 2025 demand rose from 298 MW at 04:00 to 450 MW at 16:00; on 21 January it had two humps, at 07:00 and 19:00.
Weekly rhythm
Weekend hours ran 4% below weekdays on average in 2025, and up to 9% lower around 07:00.
Seasons
July held the top hour; winter cold snaps made a second, lower season; April was the quietest month.
Load-duration curve
Sorted from highest to lowest, only 180 of 8,760 hours needed more than 90% of the peak.
Duck curve
Solar lowers midday net demand and steepens the evening climb that other plants must meet.
Go deeper · AnalystLoad factor, and the area under the curve

The load factor is average demand \(\bar d\) over peak demand \(d_{\max}\):

\[LF = \frac{\bar d}{d_{\max}} = \frac{282\ \text{MW}}{450\ \text{MW}} = 0.63 .\]

Each hour adds its demand times one hour of energy. Sorting changes the order, not the total: \(E = \sum_h d_h \times 1\ \text{h} \approx 2{,}469{,}131\) MWh, about 2,469 GWh, either way. Here \(d_h\) is demand in hour \(h\), in MW, and \(\sum_h\) adds all 8,760 hours.

The energy above a line \(L\) counts only the part of each hour’s demand over the line, \(\max(0,\ d_h - L) \times 1\ \text{h}\), in MWh. For \(L = 405\) MW it is 3,068 MWh, only 0.12% of \(E\): the top tenth of the peak holds very little of the year’s energy.

Go deeper · ModelerWhy models keep the hours in order
  • Sorting keeps how many hours each level of demand is needed.
  • Sorting loses which hour follows which. A battery’s charge, a plant’s ramp limit and the duck’s evening climb all depend on that order.
  • Models with memory keep enough of the order, in different ways, each with its own limits.
  • Next: lesson 9 lets storage fill the duck’s belly; lesson 11 solves a day hour by hour with a ramp rule; lesson 13 plans a year from representative weeks linked in calendar order.
Go deeperWhat the data leave out
  • One region only: the Southwest Power Pool in the central U.S., scaled down. Other regions peak at other hours and in other seasons.
  • For 699 hours of 2025 (about 8%) SPP's reported demand was implausible; those hours were rebuilt from SPP's own day-ahead forecast. The named days contain no rebuilt hours.
  • SPP's solar telemetry froze for 1,377 hours (about 16%); those hours were filled from the same hour on nearby days, and solar is set to zero after dark. The named days contain no filled hours.
  • SPP has little solar (0.8% of its 2025 generation), so the duck curve adds a hypothetical farm that follows SPP's solar shape.
  • The solar slider sets an illustrative maximum output, the farm's output in SPP's top solar hour of 2025; it is not a nameplate rating.
  • Real net demand subtracts wind as well as solar; here only solar is subtracted, to isolate its effect.
  • One year of data cannot show long-term growth.

Check yourself

Myth or fact? “Demand peaks at noon, when the sun is strongest.”

  • Myth
  • Fact
Answer

Myth. On 29 July 2025 the peak came at 16:00 to 17:00. Hot-day peaks often come after noon, as buildings hold heat and people come home.

Myth or fact? “Power plants are built for average demand.”

  • Myth
  • Fact
Answer

Myth. The grid as a whole must be able to meet the peak hour, with a margin: Prairie Town averages 282 MW but peaks at 450 MW.

Myth or fact? “Adding solar makes the evening easier for the other plants.”

  • Myth
  • Fact
Answer

Myth. Solar fades while demand stays high, so the others must climb faster: on 30 May 2025, with 200 MW of solar, net demand's steepest three-hour climb after noon is 114 MW, against 20 MW without it.

180 of the year's 8,760 hours need more than 405 MW. What share of the year is that?

  • 21%
  • 2.1%
  • 0.2%
  • 18%
Answer

2.1%: 180 ÷ 8,760 = 0.021. Some resources are needed only in hours like these; how often each actually runs depends on the rest of the system.

Which change can raise Prairie Town's winter electricity demand by moving heating from gas to electricity?

  • More solar panels
  • Electric heat pumps replacing gas furnaces
  • LED lighting everywhere
  • A four-day work week
Answer

Heat pumps. A heat pump uses electricity to move heat, so heating moves from gas onto the grid. How much demand rises depends on adoption, weather and equipment.

Numbers, sources and what’s next

Numbers to remember

Number What it means Source
54.4 GW The Southwest Power Pool's busiest hour of 2025: 16:00 to 17:00 on 29 July. EIA-930 six-month files, computed · data for 2025
299 TWh SPP's demand over 2025: an average of 34 GW, so a load factor of 0.63. EIA-930 six-month files, computed · data for 2025
759 GW Record peak demand of the Lower 48 states, on the same day, 29 July 2025. EIA, Today in Energy · as of August 2025
180 hours Hours of 2025 with demand above 90% of the year's peak (SPP's shape): 2.1% of the year. EIA-930 six-month files, computed · data for 2025
0.8% Solar's share of SPP's net generation in 2025. EIA-930 six-month files, computed · data for 2025

Next on the starter path

Lesson 3 · How does the grid stay balanced every second?

In the real world

Every U.S. balancing authority reports its demand each hour; EIA's Hourly Electric Grid Monitor draws yesterday's curve for your region.

In California, where solar is a large share, EIA reports the duck curve getting deeper as solar capacity grows (2023).

Sources and further reading

  • EIA. Hourly Electric Grid Monitor (Form EIA-930), and its six-month files: the Southwest Power Pool's 2025 data behind every curve here.
  • EIA (2025). U.S. electricity peak demand set new records twice in July. Today in Energy.
  • EIA (2023). As solar capacity grows, duck curves are getting deeper in California. Today in Energy.
  • EIA. Electricity explained: Use of electricity (2025 data).

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.

© 2026 Honglin Li · Kansas State University · hlli@ksu.edu

 

Last updated: November 2026 · Built with Quarto