Definition
Peak Shaving and Valley Filling (Peak Shaving) refers to a power management strategy that smooths the load curve by reducing the peak load on the power grid and filling the low load valleys through technical means. It uses energy storage systems, demand-side response, adjustable loads, and other methods to decrease electricity drawn from the grid during peak periods and increase electricity use or discharge to the grid during off-peak periods, thereby balancing supply and demand, alleviating grid pressure, and reducing electricity costs for users.
Technical methods of peak shaving and valley filling
Energy Storage Peak Shaving and Valley Filling: Energy storage systems charge during off-peak periods and discharge during peak periods, directly reducing the user's peak electricity demand from the grid. Load Shifting: Using demand-side response, adjustable loads such as air conditioners and production equipment are shifted from peak periods to off-peak periods. Distributed Photovoltaics: Distributed photovoltaics generate electricity during the day, helping to relieve midday peaks; if combined with energy storage, the curve can be further smoothed. Microgrids: An energy microgrid coordinates source, grid, load, and storage so the local peak can be cut and the valley filled. These methods are often used in combination and require strategies based on load forecasting and smart meter data.
Relationship with demand charges and power quality
For commercial and industrial users on a two-part tariff, the demand charge is billed on maximum demand in kilowatts, so cutting the peak can cut that charge by a large amount. Power quality and supply reliability still have to hold while the peak is shaved and the valley is filled. The storage system needs overload protection and harmonic suppression, and load changes must not disturb critical equipment. Power quality monitoring watches voltage, current, and power factor in real time so the site stays within safe electrical limits.
Application of the Internet of Things in peak shaving and valley filling
The Internet of Things supplies the measurements, carries out the strategy, and checks the result. Smart meters and power quality monitors collect live load and electrical data and send them to a cloud platform. Load forecasting and artificial intelligence then produce a charge, discharge, and load-shift plan. An edge controller issues the storage charge and discharge commands and starts or stops air conditioners. SimtoLink can provide device connectivity, the data path, remote control, and API integration for this use, so the site can reshape its load and reduce cost.
Application Scenarios and Selection Recommendations
Peak shaving and valley filling fit sites with a high demand charge, including data centers, industrial parks, charging stations, and cold-chain cold stores. Selection should weigh the present load curve and the peak-to-valley gap, storage capacity and charge/discharge C-rate, the demand charge and the peak-valley tariff, the power quality requirement, and the payback period. As storage costs fall and electricity markets open further, the economics of shifting load between peak and valley improve sharply.
Electrical Constraints on Peak and Valley Shifting
Peak shaving and valley filling change when energy is taken from the public grid: less at the peak, more in the valley, or charging storage then, so that billed demand or the system peak falls. It does not make total consumption disappear. When the shift fails, daily energy may be almost unchanged while maximum demand does not fall at all. Evaluation has to look at when maximum demand occurs and how long it lasts, not only at whether the curve looks flatter.
It is feasible only with something truly controlled, such as a line that can pause, cooling that can be finished early, or storage allowed to charge in the valley. Critical load that cannot stop should not be written onto the adjustable list. The strategy must finish before the demand window closes. Discharge after the window does not help maximum demand that has already formed. The point-of-connection capacity, discharge rate, and protection settings must cover the planned power. Otherwise the equipment trips first, and the curve shows a deeper gap at the trip.
It is easy to treat a cleaner display as a lower load. Stopping acquisition upstream of the metering point flattens the curve while the real load remains. If a large photovoltaic output is mistaken for successful peak shaving because purchases fell, demand reappears in the evening when the photovoltaic output drops. If storage is already full in the valley and still charges on plan, there is no reserve when the real peak arrives. If the forecast places the peak too early, the battery is empty ahead of time.
Regulation also has to hold voltage and protection, not only the power number. If the smart grid side issues demand response at the same time, the local strategy must know which command has priority, so two controllers do not drive the same feeder in opposite directions. Machine learning can indicate which period tomorrow's peak falls in, but charge and discharge lockouts, minimum duration, and maintenance state are still set by electrical constraints. A score cannot close a breaker by itself.
How demand is counted decides whether peak shaving counts as success. Some contracts take the maximum of fifteen-minute average power. Others use a sliding fixed window. Clipping one spike in instantaneous power does not change the bill if the window average does not fall. The strategy should be designed to the statistical method in the contract, not to the second-by-second curve on the screen. Discharge that starts only after the window has opened can affect only the time left in that window. Power already counted cannot be pulled back.
Adjustable load should be written as a list: which unit can stop, the shortest stop, which process it affects, and how long warmup takes to restart. Without that list, optimization can only assume "every load can move," and the first attempt on site hits a furnace that cannot stop. Storage should also keep an emergency reserve. If all of the energy is used for peak shaving, there is nothing left to support critical load when the grid is disturbed.
A review should compare maximum demand at the same metering point and the same statistical window, and note whether that day had maintenance, rationing, or unusual weather. Laying two days' curves on top of each other does not show whether the strategy worked or the production plan was simply light. If inverters, air conditioners, and chargers each have their own controller, they must share a clock. Otherwise one is still charging while another has started discharging, and power at the point of connection does not fall.