A city does not use energy at one constant rate. Demand rises when people wake, trains accelerate, offices open, factories start, kitchens cook and air-conditioning fights afternoon heat. It falls when buildings empty and machines stop. The resulting pattern is a load profile: a time-based picture of how much power a system asks for.
Energy demand describes how much energy is used over a period; load describes the rate at which energy is being used at a particular time. The distinction is crucial. Two buildings can consume the same monthly kilowatt-hours but create very different stress on the grid if one concentrates demand into sharp peaks while the other spreads it evenly.
Wait, what? The same amount of energy can require a very different power system
Suppose two facilities each consume 2,400 kWh in a day. One draws a steady 100 kW for 24 hours. The other consumes almost everything during a four-hour period, reaching several hundred kilowatts. Their daily energy is the same, but the second facility needs larger cables, transformers, switchgear and supply capacity to serve the peak.
This is why electricity systems care about both how much energy is used and when it is used.
The direct answer
Energy demand and load profiles work by mapping power use across time. Operators and planners study the baseline, daily peaks, seasonal changes, sudden ramps, minimum demand, coincidence between users and the duration of high-load periods. Those shapes determine how much generation, network capacity, storage, reserve and flexibility the system needs.
The basic chain is:
- people and machines request useful services,
- end-use devices convert energy at changing rates,
- individual loads combine into a system profile,
- the grid must match that combined power demand continuously,
- storage, flexible demand and generation reshape the residual profile.
Energy versus load
Energy is power integrated over time. A 1 kW appliance running for one hour uses 1 kWh. A 2 kW appliance running for half an hour also uses 1 kWh. The energy total is identical, but the instantaneous load is different.
Infrastructure must be capable of handling the highest relevant load, not only the average. This is why peak demand can drive investment even when the peak occurs for only a few hours each year.
Baseload and minimum demand
Every system has a level of demand that persists for much of the time. Historically this was often described as baseload. In modern grids, it is often more useful to talk about minimum demand and the shape of residual demand after variable generation is included.
Some loads truly are persistent: data centres, refrigeration, communications, hospital systems, standby equipment and industrial processes. Others vary strongly with occupancy, weather or schedule.
Peak demand
Peak demand is the highest power drawn during a defined interval. The peak can be system-wide or local. A national grid may experience a seasonal peak. A building may peak in the afternoon. A neighbourhood transformer may peak in the evening when residents return home.
The location matters because network constraints are local. A national system can have spare generation while one distribution feeder is overloaded.
Load factor
Load factor compares average demand with peak demand over a period. A high load factor means demand is relatively even. A low load factor means the system has sharp peaks relative to average use.
High load factor can improve utilisation of infrastructure because expensive equipment is used more consistently. But an extremely flat load is not inherently desirable if it comes from unnecessary consumption. The goal is efficient service delivery, not maximising equipment usage for its own sake.
Coincidence: individual peaks do not always happen together
If ten homes each have a 10 kW maximum possible load, the neighbourhood does not automatically require 100 kW because not every appliance in every home operates simultaneously. Engineers use diversity and coincidence concepts to estimate how individual loads combine.
Large systems benefit from statistical diversity, but electrification can change coincidence patterns. If many vehicles begin charging at exactly the same evening time, diversity can shrink and local peaks can become sharper.
Ramps: how quickly demand changes
Maximum load is not the only challenge. Systems must also respond to ramps—rapid changes in demand or net demand. A railway timetable can create repeated traction ramps. A large industrial facility can switch major equipment on or off. Evening lighting and cooling patterns can change demand quickly.
Generation and storage need enough ramp capability to follow these changes while keeping frequency and voltage within limits.
Net load
Net load is the demand remaining after selected generation—often variable renewable generation such as solar or wind—is subtracted. A region can have high gross electricity use but relatively low net grid demand during sunny hours if rooftop solar supplies much of the load locally.
As solar grows, the net-load curve can fall strongly at midday and rise rapidly toward evening as sunlight decreases while consumer demand remains. This changes the ramping problem even if total daily energy demand changes little.
The duck curve
In systems with substantial solar generation, midday net demand can become low while evening net demand rises steeply after sunset. Plot the shape and it can resemble a duck, giving rise to the term “duck curve”.
The important idea is not the animal-shaped graph. It is the operational consequence: the grid needs enough flexible resources to reduce output during the solar-rich period and increase output rapidly later.
Weather changes load
Temperature, humidity, sunlight, wind and rainfall can all affect energy demand. In warm climates, hot humid afternoons increase air-conditioning and dehumidification load. Cloud cover can reduce solar generation while changing cooling demand. Storms can alter both supply and consumption.
Forecasting therefore combines historical demand with weather models, calendars, economic activity and real-time data.
Daily cycles
Human schedules create repeated daily structure. Residential demand may rise in mornings and evenings. Commercial buildings rise after opening and fall after closing. Transport loads follow commute patterns. Industrial demand may be flatter or shift-based.
Weekends and public holidays change the profile because occupancy and industrial schedules change. A good forecast recognises calendar structure rather than treating every day as identical.
Seasonal load
In temperate climates, electricity demand may peak in winter because of heating or summer because of cooling. In tropical climates, seasonal changes can be smaller but weather, school schedules, tourism, rainfall and economic activity still alter demand.
Long-term infrastructure planning therefore uses multi-year demand profiles rather than one representative day.
Buildings as load profiles
A building is not one load. It is a stack of cooling, ventilation, lighting, lifts, plug loads, pumps, hot water, kitchens, servers and specialist equipment. Each has its own schedule and power characteristics.
Building-management systems can reveal which loads create the peak. Sometimes a small number of short-duration systems determine the maximum demand charge or electrical infrastructure size.
Cooling and afternoon peaks
Cooling demand is driven by solar gain, outdoor temperature and humidity, internal equipment, people and ventilation. Thermal mass can delay the peak because walls and furnishings absorb heat before it reaches the cooling system.
Pre-cooling and chilled-water storage can shift some electricity use earlier, reducing the afternoon electrical peak without reducing comfort.
Transport load profiles
Electric rail systems draw high power when trains accelerate and can return power during regenerative braking. The system load therefore contains repeated pulses shaped by timetable, train position, gradients and passenger demand.
Electric-vehicle charging creates a different profile. Home charging may cluster in the evening. Depot fleets can create large predictable blocks. Fast-charging stations can create sharp local peaks. Smart charging can shift some of this demand to better times.
Industrial load profiles
Factories may have large motors, furnaces, compressors, pumps and process equipment. Some processes operate continuously while others run in batches. Motor starting can create brief high-power events. Electric furnaces can create major block loads.
Industrial demand can sometimes be flexible if production stages can shift in time, but process constraints, labour, quality and safety determine how much shifting is realistic.
Data centres
Data centres can create large relatively continuous electrical loads. Computing demand varies, but facilities often maintain substantial baseline power for servers, networking, storage and cooling.
Workload scheduling can provide some flexibility where applications allow it. Cooling and backup systems also influence the profile. As data-centre clusters grow, their local network impact can become a major planning issue.
Demand response
Demand response changes electricity use in response to system conditions, prices or instructions. A building can adjust cooling slightly. A factory can postpone a process. A fleet can delay charging. A water utility can shift pumping within reservoir constraints.
The key is to preserve the final service while changing timing. Flexibility is most valuable when it reduces peaks, fills low-demand periods or responds rapidly to unexpected imbalance.
Peak shaving
Peak shaving reduces the highest load. Batteries can discharge during peaks. Thermal storage can reduce chiller power. Backup generation may operate in permitted contexts. Controls can stagger equipment startup.
Reducing a short peak can avoid or delay upgrades to transformers, cables and generation capacity even if total energy consumption changes only modestly.
Load shifting
Load shifting moves energy use from one time to another. Charging a battery overnight and discharging during an afternoon peak is one example. Making ice or chilled water during a low-demand period and using it later for cooling is another.
Load shifting does not necessarily reduce total energy use. Storage losses can even increase gross consumption. The value comes from improving timing and reducing stress on constrained infrastructure.
Valley filling
Low-demand periods can leave generation or networks underused. Flexible demand can move into these valleys. Electric-vehicle charging is a common candidate because vehicles may be parked for many hours but need only a fraction of that time to charge.
Valley filling can improve asset utilisation and absorb surplus low-cost generation, provided it does not create a new peak.
Storage changes the load seen by the grid
Storage can charge when demand is low or generation is abundant and discharge when demand is high. From the grid’s perspective, charging increases load and discharging reduces net load.
This makes storage a profile-shaping resource. But state of charge creates a memory: storage cannot discharge indefinitely and must be recharged later. A control strategy must therefore optimise over time, not one instant.
Electricity prices can follow load shape
Wholesale electricity prices often rise when demand is high relative to available low-cost supply and fall when abundant generation is available. Time-of-use tariffs can expose consumers to simplified versions of these system conditions.
Price is not a physical quantity, but it can become a control signal that changes physical demand. Economics and engineering meet through the load curve.
Demand charges
Some commercial electricity tariffs include charges based on maximum demand rather than only total kilowatt-hours. This reflects the cost of infrastructure that must be sized to serve peaks.
A facility can therefore save money by reducing a short peak even if total monthly energy use barely changes. This creates an economic incentive for load management.
Forecasting demand
Operators forecast demand from recent measurements, time of day, weather, calendar events, economic conditions and known large loads. Short-term forecasts support dispatch. Medium-term forecasts support maintenance and fuel planning. Long-term forecasts shape investment in power plants, grids and storage.
Forecast error matters because electricity must be balanced in real time. Reserves absorb the residual between prediction and reality.
Electrification reshapes demand
Electric vehicles, heat pumps, electric industrial processes and data centres can raise electricity demand while reducing direct fossil-fuel use elsewhere. The important planning question is not only annual energy growth but the timing of the new loads.
Managed charging and thermal flexibility can make electrification easier to integrate. Unmanaged clustering can create new local peaks even when the national system has enough annual energy.
Energy efficiency changes the load profile too
Efficiency can lower total consumption and reduce peaks, but the effect depends on which end use is improved. More efficient air-conditioning can cut afternoon peak demand strongly. Efficient overnight lighting may reduce energy without affecting the system peak much.
This is why planners value when savings occur, not only how many kilowatt-hours are saved.
Distributed solar reshapes demand seen by the grid
Rooftop solar can supply local loads behind the meter. The building still consumes electricity internally, but less is drawn from the grid during sunny periods. Grid operators see a modified net-load curve.
If solar output falls rapidly because of weather or sunset, grid demand can rise even if the building’s underlying service demand has barely changed. The distinction between gross load and net load becomes important.
Local congestion
A grid can have enough total generation and still face local congestion if transmission or distribution paths cannot carry the required power. New electric-vehicle clusters, industrial loads or data centres can strain a particular substation before national supply becomes scarce.
Load profiles are therefore spatial as well as temporal. Engineers need to know where and when power is required.
Coincident renewable output and demand
A renewable generator is especially valuable when its output coincides with demand that would otherwise be difficult or expensive to serve. Solar can align well with daytime cooling in some regions. Wind may follow different patterns.
Annual energy yield alone therefore does not determine system value. Timing and location affect how much generation displaces other capacity or reduces network load.
Load duration curves
A chronological load profile shows when demand occurs. A load duration curve rearranges those observations from highest to lowest load. This reveals how many hours the system operates above any given demand level.
The curve helps distinguish capacity needed only for rare peaks from capacity used for much of the year. It is useful for planning, though it loses chronological information about ramps and sequence.
Capacity planning
Generation and network capacity must cover expected demand plus uncertainty, maintenance and contingencies. Planning uses peak forecasts, resource availability, reserve requirements and transmission constraints.
A megawatt of installed capacity is not automatically a megawatt of dependable capacity at the system peak. Weather-dependent resources, outages and fuel constraints affect availability.
Singapore and load shape
Singapore’s warm climate, dense commercial activity, industry, rail transport, data infrastructure and cooling demand create a distinctive electricity profile. Solar generation can reduce daytime grid demand, while cooling and commercial loads keep daytime power needs significant.
As electrification, data centres, storage and regional interconnection evolve, the important question is not only how much electricity Singapore uses in a year but how the hourly and local shape changes. The detailed market and system-operation mechanisms remain with dedicated Singapore and grid pages; this article owns the general load-profile logic.
Three worked reasoning examples
1. Two buildings with equal monthly energy
Building A uses moderate power throughout the day. Building B has a sharp two-hour afternoon peak because several chillers and lifts operate together. Monthly kWh can be similar, but Building B needs larger peak capacity. Staggering loads or using thermal storage can reduce the peak without changing the service delivered.
2. Evening electric-vehicle charging
Many vehicles arrive home at similar times and begin charging immediately. Residential demand and vehicle charging coincide, creating a local peak. Smart charging delays some vehicles until later while ensuring each reaches its required state of charge by morning. Total energy may remain nearly unchanged while peak power falls.
3. Midday solar plus evening demand
Solar generation pushes net grid demand down at midday. As the Sun sets, output falls quickly while homes and businesses still consume electricity. The system needs resources that can ramp upward. Batteries, flexible loads, imports or generators can fill the gap. The challenge is the shape, not the daily energy total alone.
Common misconceptions
- Energy consumption and peak load are not the same quantity.
- The national peak and a local feeder peak can occur at different times.
- Baseload is not a requirement that one specific technology must run continuously.
- Storage can shift load but must recharge later.
- Demand response does not necessarily mean reducing service; it can shift timing.
- Annual renewable generation does not reveal whether output coincides with peak demand.
- Average demand can hide infrastructure-driving peaks and ramps.
A universal load-profile audit
- Measure demand at a time resolution appropriate to the system.
- Calculate total energy use.
- Identify peak demand and its duration.
- Identify minimum load and persistent baseline loads.
- Measure ramp rates.
- Check which loads coincide to create the peak.
- Separate gross load from net load after local generation.
- Locate spatial bottlenecks in feeders, transformers or substations.
- Test demand response, storage and efficiency options.
- Forecast how electrification and new loads will change the future profile.
How load profiles fit the wider Energy series
Load profiles connect power, storage, resilience and electricity grids. They explain why an energy system must be designed for time, not just totals.
The deeper lesson is that civilisation does not consume energy as one annual number. It asks for power moment by moment, and the shape of that request determines the machines, networks and buffers required to keep everything working.
How Energy Works | Main Series
How Power Works · How Energy Storage Works · How Energy Security and Resilience Work