In short. Load shedding disconnects less important loads so the available sources can carry the important ones. It is needed when generator capacity is limited, when a unit fails, or when frequency or voltage begins to sag. A good scheme has a clear priority list, fast and correct triggers, and a planned way to bring loads back.
Why shed load
- A generator or the available generators cannot carry the full load.
- One generator fails in a multi-generator plant and the rest must carry the load.
- Frequency or voltage falls during a step or an overload.
- Utility returns weak, or a fuel limit means reduced operation.
Priority list
| Priority | Load type | Action |
|---|---|---|
| 1 | Life-safety and mission-critical | Never shed |
| 2 | Essential process and services | Shed last |
| 3 | Important but deferrable | Shed after lower tiers |
| 4 | Comfort and non-essential | Shed first |
Build the list from the classification in critical, essential and mission-critical power.
Trigger types
- Static (capacity-based): shed according to the number of generators online versus load. Fast and predictable.
- Dynamic (frequency-based): shed when frequency falls below set points, with rate-of-change checks. Catches unexpected events.
- Overload and power-based: shed when a unit’s loading exceeds a threshold for a set time.
- Event-based: shed immediately when a known event occurs, such as loss of a generator.
Timing
Shedding for a generator trip must be fast, within tens to hundreds of milliseconds in many cases, or the remaining generators may stall. Dedicated relays or PLCs with direct wiring are often used for the fast path. See load-shedding automation.
Settings to avoid nuisance shedding
- Ignore short transients such as motor starts, using time delays or rate-of-change logic.
- Stage the shedding, one tier at a time, with checks that it worked before shedding more.
- Latch the shedding until an operator or a restoration sequence resets it.
Restoration
Bring loads back one tier at a time after the generators are stable and capacity is confirmed. Do not restore everything at once: the combined step may cause a second collapse. See restoration and restart sequencing.
Worked illustration
A plant has three 1,000 kW generators and 2,200 kW of load. Capacity with three units is 3,000 kW, enough. If one generator trips, 2,000 kW remains against 2,200 kW load, so 200 kW or more must go. The static scheme knows which tier to shed at once, say 300 kW of comfort and deferrable load, and the generators stay stable. Without it, the two units overload, frequency drops and the plant may trip into a full blackout.
Testing
- Simulate each generator trip and confirm shedding and the system response.
- Record the time from event to breaker open.
- Test the restoration sequence.
- Re-test after any change to loads or settings.
When capacity falls short
Events such as a failed generator, a degraded UPS or a cooling loss leave less capacity than load. Without a plan, the whole bus fails. Load shedding drops the least important loads first to protect the most important.
| Tier | Examples | Action |
|---|---|---|
| 1 (protect always) | Life safety, control, IT core, monitoring | Never shed |
| 2 | Production-critical, key cooling | Shed only in severe shortage |
| 3 | Important but deferrable | Shed in shortage |
| 4 | Comfort and convenience | Shed first |
Triggers
- Loss of a source or breaker status change.
- Available capacity below load.
- Frequency falling below a threshold.
- Battery state of charge low.
- High temperature in cooling or electrical room.
Calculation example (illustrative)
Capacity online 1,500 kW (three 500 kW sets). Safe limit 90 percent: 1,350 kW. Load 1,600 kW before an event. After one set trips, capacity 1,000 kW × 0.9 = 900 kW. Shed 700 kW. Tier 4 and 3 loads total 650 kW, so a part of tier 2 is also required.
Do not forget
- Cooling for critical equipment must not be shed before the heat load is.
- Inform operators and log all shedding.
See load shedding automation and load shedding engineering.
Common mistakes
- Priority lists that are out of date.
- Slow shedding relying on a supervisory system.
- Restoring all loads at once.
What to send us
Send the load list, generator data and the events to cover. Send your application details. See also load-shedding planning.
Questions engineers ask
What is the difference between static and dynamic shedding?
Static shedding acts on known capacity and events. Dynamic shedding reacts to measured frequency, catching unforeseen conditions.
How fast must shedding be?
For generator loss it must act before the remaining units stall. That is often within a fraction of a second, so dedicated fast logic is used.
Can shedding be fully automatic?
Usually yes for the first response. Restoration may be automatic or operator-supervised.
How often should the priority list be reviewed?
Whenever loads change, and at least periodically as part of design review.
This guide gives general principles. Settings must be developed and tested for the actual system.