# Peak Demand Management

> Changing the shape of a load curve rather than its area — and knowing which levers move it.

Peak demand management means changing the shape of a load curve, not its area. That is a different problem from energy efficiency, and it is often cheaper, because shifting a load costs less than eliminating it. This section covers the practical levers — sequencing and staggered startup, thermal storage, batteries, on-site generation, curtailment agreements, control settings and interval-data analysis — with the arithmetic for deciding which of them a site actually needs. It also covers the ones that fail: measures that cut kilowatt-hours while leaving the peak untouched, and control strategies that shave one meter while pushing the peak onto another. Every strategy here is judged against the same test, which is whether the next twelve bills fall.

## What peak demand management means

Peak demand management is the set of measures that lower the highest fifteen-minute average your meter records in a billing period, which is the figure a demand charge is priced on. It works on when load happens rather than on how much energy is used, and that is why it is a different discipline from energy efficiency.

The names overlap. *Peak load management*, *demand charge management* and *peak shaving* describe the same goal from different sides: the first is the engineering name, the second the accounting one, the third what happens to the load curve. A vendor's choice of name says nothing about which of the methods below it is selling.

## How to manage peak demand: the methods, cheapest first

The methods below run roughly from free to capital-intensive. Which one a site needs is decided by the shape of its load, not by preference. The cheap ones come first for a second reason as well: they change the peak that any expensive measure would later be sized against.

| Method | What it changes | Cost | Tends to suit | Read |
| --- | --- | --- | --- | --- |
| Staggered startup | Machines restarting together after a break | Usually nothing | Sharp peaks at shift start | [Staggered startup](https://thedemandcharge.com/articles/staggered-startup-sequencing) |
| Load scheduling | When flexible loads run | Low | Charging, tank heating, batch work | [Load shifting](https://thedemandcharge.com/articles/load-shifting-vs-energy-efficiency) |
| Pre-cooling | When the cooling plant works hardest | Low | Afternoon cooling peaks | [HVAC scheduling](https://thedemandcharge.com/articles/hvac-scheduling-demand) |
| Demand limiting controls | Sheds ranked loads as the interval average climbs | Moderate | Sites with genuinely deferrable load | [Demand limiting](https://thedemandcharge.com/articles/demand-limiting-controls) |
| Thermal energy storage | Moves cooling production into the night | Significant capital | Large, predictable cooling peaks | [Thermal storage](https://thedemandcharge.com/articles/thermal-energy-storage-demand) |
| Battery storage | Discharges into the peak | High capital | Short, sharp, frequent peaks | [Battery sizing](https://thedemandcharge.com/articles/battery-peak-shaving-sizing) |
| Demand response | Curtails load on request, for payment | Varies | Sites with a curtailment plan | [Demand response](https://thedemandcharge.com/articles/demand-response-programs) |

## Peak load management systems

Two of those rows cannot be run by hand. A demand limiting controller, the core of any peak load management system, watches the running average inside the current interval, predicts where it will finish, and sheds pre-ranked loads before the target is passed. A battery needs the same forecast to decide when to discharge.

Both fail in predictable ways: nothing on site is really sheddable, the controller's interval is not synchronized with the utility meter's, restored loads create the next peak, or the target is set in the wrong place. Each is covered in [demand limiting controls and how they fail](https://thedemandcharge.com/articles/demand-limiting-controls), together with how to specify a controller that avoids them.

## Your own peak, or the grid's

Most demand charges bill your own maximum. Some bill what you were drawing when the regional grid peaked: ERCOT's four coincident peak and PJM's capacity tag are the best-known cases. Managing those is a forecasting exercise rather than a control one, because the interval that counts is chosen by the system and known only afterward. [Coincident vs. non-coincident peak demand](https://thedemandcharge.com/articles/coincident-vs-non-coincident-demand) explains how to tell which one you are paying.

## Where to start

With data, not equipment. A year of fifteen-minute interval data shows when the peaks happen, how often and what causes them, and the utility can usually supply it at no cost: [how to get your interval data](https://thedemandcharge.com/articles/interval-data-how-to-get-it).

The full method, from that diagnosis through the ladder of measures to the arithmetic for choosing between them, is in [how to reduce peak demand charges](https://thedemandcharge.com/articles/how-to-reduce-peak-demand-charges). Before anything is funded, it is worth knowing which measures save energy without touching the peak at all: [six efficiency measures that do not cut your demand charge](https://thedemandcharge.com/articles/measures-that-dont-cut-demand).

## Articles in this section

- [How to Reduce Peak Demand Charges](https://thedemandcharge.com/articles/how-to-reduce-peak-demand-charges): How to reduce peak demand and the demand charge it sets: six measures from free sequencing to batteries, and the interval-data diagnostic that picks the right one.
- [Demand Limiting Controls and How They Fail](https://thedemandcharge.com/articles/demand-limiting-controls): A demand controller predicts the interval average and sheds load before the target is passed. The concept is sound; the implementations fail in a small number of predictable ways.
- [Demand Response Programs — What You Are Actually Selling](https://thedemandcharge.com/articles/demand-response-programs): You are not selling electricity. You are selling a commitment to reduce load on request, and the penalty structure for failing to deliver is the part worth reading twice.
- [Load Shifting Is Not Energy Efficiency](https://thedemandcharge.com/articles/load-shifting-vs-energy-efficiency): Efficiency reduces the area under the load curve. Shifting changes its shape. They save money on different lines of the bill, and confusing them wrecks business cases.
- [Pre-Cooling and HVAC Scheduling Against the Peak](https://thedemandcharge.com/articles/hvac-scheduling-demand): Buildings store heat. Running the cooling plant harder before the expensive window and coasting through it turns that thermal mass into free storage.
- [Six Efficiency Measures That Do Not Cut Your Demand Charge](https://thedemandcharge.com/articles/measures-that-dont-cut-demand): Every one of these saves real energy. None of them reliably reduces the highest fifteen-minute interval, and projects sold on the wrong line lose credibility for the next one.
- [Sizing a Battery for Peak Shaving](https://thedemandcharge.com/articles/battery-peak-shaving-sizing): Power rating and energy capacity are two separate specifications. A battery correct on one and wrong on the other fails to shave the peak while costing the full price.
- [Staggered Startup — The Cheapest Demand Measure There Is](https://thedemandcharge.com/articles/staggered-startup-sequencing): The highest interval of the month is frequently the moment after a break, when everything restarts together. Spreading those starts over twenty minutes usually costs nothing.
- [Thermal Energy Storage for Demand Reduction](https://thedemandcharge.com/articles/thermal-energy-storage-demand): Make cooling at night, use it during the peak. The cooling load is unchanged; the electrical load that produces it moves to hours where capacity is not being priced.
- [Why Rooftop Solar Rarely Cuts a Demand Charge](https://thedemandcharge.com/articles/solar-pv-and-demand-charges): Solar cuts kilowatt-hours reliably. It cuts the monthly peak only when the sun cooperates in every interval that matters, and Berkeley Lab's modeling shows how rarely that happens.

---

Canonical URL: https://thedemandcharge.com/topics/peak-management  
Site index for agents: https://thedemandcharge.com/llms.txt